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_conversion): Treat the number
6 of units as polynomial. Choose between WIDE and NARROW based
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 (simd_clone_subparts): New function.
14 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
16 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
17 Alan Hayward <alan.hayward@arm.com>
18 David Sherwood <david.sherwood@arm.com>
20 * tree-vect-stmts.c (vectorizable_call): Treat the number of
21 vectors as polynomial. Use build_index_vector for
24 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25 Alan Hayward <alan.hayward@arm.com>
26 David Sherwood <david.sherwood@arm.com>
28 * tree-vect-stmts.c (get_load_store_type): Treat the number of
29 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
30 for variable-length vectors.
31 (vectorizable_mask_load_store): Treat the number of units as
32 polynomial, asserting that it is constant if the condition has
33 already been enforced.
34 (vectorizable_store, vectorizable_load): Likewise.
36 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
37 Alan Hayward <alan.hayward@arm.com>
38 David Sherwood <david.sherwood@arm.com>
40 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
41 of units as polynomial. Punt if we can't tell at compile time
42 which vector contains the final result.
44 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
45 Alan Hayward <alan.hayward@arm.com>
46 David Sherwood <david.sherwood@arm.com>
48 * tree-vect-loop.c (vectorizable_induction): Treat the number
49 of units as polynomial. Punt on SLP inductions. Use an integer
50 VEC_SERIES_EXPR for variable-length integer reductions. Use a
51 cast of such a series for variable-length floating-point
54 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
55 Alan Hayward <alan.hayward@arm.com>
56 David Sherwood <david.sherwood@arm.com>
58 * tree.h (build_index_vector): Declare.
59 * tree.c (build_index_vector): New function.
60 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
61 of units as polynomial, forcibly converting it to a constant if
62 vectorizable_reduction has already enforced the condition.
63 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
64 to create a {1,2,3,...} vector.
65 (vectorizable_reduction): Treat the number of units as polynomial.
66 Choose vectype_in based on the largest scalar element size rather
67 than the smallest number of units. Enforce the restrictions
70 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
71 Alan Hayward <alan.hayward@arm.com>
72 David Sherwood <david.sherwood@arm.com>
74 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
75 number of units as polynomial.
77 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
78 Alan Hayward <alan.hayward@arm.com>
79 David Sherwood <david.sherwood@arm.com>
81 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
82 * target.def (autovectorize_vector_sizes): Return the vector sizes
83 by pointer, using vector_sizes rather than a bitmask.
84 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
85 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
86 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
88 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
89 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
90 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
91 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
92 * omp-general.c (omp_max_vf): Likewise.
93 * omp-low.c (omp_clause_aligned_alignment): Likewise.
94 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
95 * tree-vect-loop.c (vect_analyze_loop): Likewise.
96 * tree-vect-slp.c (vect_slp_bb): Likewise.
97 * doc/tm.texi: Regenerate.
98 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
100 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
101 the vector size as a poly_uint64 rather than an unsigned int.
102 (current_vector_size): Change from an unsigned int to a poly_uint64.
103 (get_vectype_for_scalar_type): Update accordingly.
104 * tree.h (build_truth_vector_type): Take the size and number of
105 units as a poly_uint64 rather than an unsigned int.
106 (build_vector_type): Add a temporary overload that takes
107 the number of units as a poly_uint64 rather than an unsigned int.
108 * tree.c (make_vector_type): Likewise.
109 (build_truth_vector_type): Take the number of units as a poly_uint64
110 rather than an unsigned int.
112 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
113 Alan Hayward <alan.hayward@arm.com>
114 David Sherwood <david.sherwood@arm.com>
116 * target.def (get_mask_mode): Take the number of units and length
117 as poly_uint64s rather than unsigned ints.
118 * targhooks.h (default_get_mask_mode): Update accordingly.
119 * targhooks.c (default_get_mask_mode): Likewise.
120 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
121 * doc/tm.texi: Regenerate.
123 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
124 Alan Hayward <alan.hayward@arm.com>
125 David Sherwood <david.sherwood@arm.com>
127 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
128 * omp-general.c (omp_max_vf): Likewise.
129 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
130 (expand_omp_simd): Handle polynomial safelen.
131 * omp-low.c (omplow_simd_context): Add a default constructor.
132 (omplow_simd_context::max_vf): Change from int to poly_uint64.
133 (lower_rec_simd_input_clauses): Update accordingly.
134 (lower_rec_input_clauses): Likewise.
136 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
137 Alan Hayward <alan.hayward@arm.com>
138 David Sherwood <david.sherwood@arm.com>
140 * tree-vectorizer.h (vect_nunits_for_cost): New function.
141 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
142 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
143 (vect_analyze_slp_cost): Likewise.
144 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
145 (vect_model_load_cost): Likewise.
147 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
148 Alan Hayward <alan.hayward@arm.com>
149 David Sherwood <david.sherwood@arm.com>
151 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
152 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
153 from an unsigned int * to a poly_uint64_pod *.
154 (calculate_unrolling_factor): New function.
155 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
157 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
158 Alan Hayward <alan.hayward@arm.com>
159 David Sherwood <david.sherwood@arm.com>
161 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
162 from an unsigned int to a poly_uint64.
163 (_loop_vec_info::slp_unrolling_factor): Likewise.
164 (_loop_vec_info::vectorization_factor): Change from an int
166 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
167 (vect_get_num_vectors): New function.
168 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
169 (vect_get_num_copies): Use vect_get_num_vectors.
170 (vect_analyze_data_ref_dependences): Change max_vf from an int *
171 to an unsigned int *.
172 (vect_analyze_data_refs): Change min_vf from an int * to a
174 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
175 than an unsigned HOST_WIDE_INT.
176 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
177 (vect_analyze_data_ref_dependence): Change max_vf from an int *
178 to an unsigned int *.
179 (vect_analyze_data_ref_dependences): Likewise.
180 (vect_compute_data_ref_alignment): Handle polynomial vf.
181 (vect_enhance_data_refs_alignment): Likewise.
182 (vect_prune_runtime_alias_test_list): Likewise.
183 (vect_shift_permute_load_chain): Likewise.
184 (vect_supportable_dr_alignment): Likewise.
185 (dependence_distance_ge_vf): Take the vectorization factor as a
186 poly_uint64 rather than an unsigned HOST_WIDE_INT.
187 (vect_analyze_data_refs): Change min_vf from an int * to a
189 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
190 vfm1 as a poly_uint64 rather than an int. Make the same change
191 for the returned bound_scalar.
192 (vect_gen_vector_loop_niters): Handle polynomial vf.
193 (vect_do_peeling): Likewise. Update call to
194 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
195 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
197 * tree-vect-loop.c (vect_determine_vectorization_factor)
198 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
199 (vect_get_known_peeling_cost): Likewise.
200 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
201 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
202 (vect_transform_loop): Likewise. Use the lowest possible VF when
203 updating the upper bounds of the loop.
204 (vect_min_worthwhile_factor): Make static. Return an unsigned int
206 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
207 polynomial unroll factors.
208 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
209 (vect_make_slp_decision): Likewise.
210 (vect_supported_load_permutation_p): Likewise, and polynomial
212 (vect_analyze_slp_cost): Handle polynomial vf.
213 (vect_slp_analyze_node_operations): Likewise.
214 (vect_slp_analyze_bb_1): Likewise.
215 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
216 than an unsigned HOST_WIDE_INT.
217 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
218 (vectorizable_load): Handle polynomial vf.
219 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
221 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
223 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
224 Alan Hayward <alan.hayward@arm.com>
225 David Sherwood <david.sherwood@arm.com>
227 * match.pd: Handle bit operations involving three constants
228 and try to fold one pair.
230 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
232 * tree-vect-loop-manip.c: Include gimple-fold.h.
233 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
234 niters_maybe_zero parameters. Handle other cases besides a step of 1.
235 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
236 Add a path that uses a step of VF instead of 1, but disable it
238 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
239 and niters_no_overflow parameters. Update calls to
240 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
241 Create a new SSA name if the latter choses to use a ste other
242 than zero, and return it via niters_vector_mult_vf_var.
243 * tree-vect-loop.c (vect_transform_loop): Update calls to
244 vect_do_peeling, vect_gen_vector_loop_niters and
245 slpeel_make_loop_iterate_ntimes.
246 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
247 (vect_gen_vector_loop_niters): Update declarations after above changes.
249 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
251 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
252 128-bit round to integer instructions.
253 (ceil<mode>2): Likewise.
254 (btrunc<mode>2): Likewise.
255 (round<mode>2): Likewise.
257 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
259 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
260 unaligned VSX load/store on P8/P9.
261 (expand_block_clear): Allow the use of unaligned VSX
264 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
266 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
268 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
269 swap associated with both a load and a store.
271 2018-01-02 Andrew Waterman <andrew@sifive.com>
273 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
274 * config/riscv/riscv.md (clear_cache): Use it.
276 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
278 * web.c: Remove out-of-date comment.
280 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
282 * expr.c (fixup_args_size_notes): Check that any existing
283 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
284 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
285 (emit_single_push_insn): ...here.
287 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
289 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
290 (const_vector_encoded_nelts): New function.
291 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
292 (const_vector_int_elt, const_vector_elt): Declare.
293 * emit-rtl.c (const_vector_int_elt_1): New function.
294 (const_vector_elt): Likewise.
295 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
298 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
300 * expr.c: Include rtx-vector-builder.h.
301 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
302 directly on the tree encoding.
303 (const_vector_from_tree): Likewise.
304 * optabs.c: Include rtx-vector-builder.h.
305 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
306 sequence of "u" values.
307 * vec-perm-indices.c: Include rtx-vector-builder.h.
308 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
309 directly on the vec_perm_indices encoding.
311 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
313 * doc/rtl.texi (const_vector): Describe new encoding scheme.
314 * Makefile.in (OBJS): Add rtx-vector-builder.o.
315 * rtx-vector-builder.h: New file.
316 * rtx-vector-builder.c: Likewise.
317 * rtl.h (rtx_def::u2): Add a const_vector field.
318 (CONST_VECTOR_NPATTERNS): New macro.
319 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
320 (CONST_VECTOR_DUPLICATE_P): Likewise.
321 (CONST_VECTOR_STEPPED_P): Likewise.
322 (CONST_VECTOR_ENCODED_ELT): Likewise.
323 (const_vec_duplicate_p): Check for a duplicated vector encoding.
324 (unwrap_const_vec_duplicate): Likewise.
325 (const_vec_series_p): Check for a non-duplicated vector encoding.
326 Say that the function only returns true for integer vectors.
327 * emit-rtl.c: Include rtx-vector-builder.h.
328 (gen_const_vec_duplicate_1): Delete.
329 (gen_const_vector): Call gen_const_vec_duplicate instead of
330 gen_const_vec_duplicate_1.
331 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
332 (gen_const_vec_duplicate): Use rtx_vector_builder.
333 (gen_const_vec_series): Likewise.
334 (gen_rtx_CONST_VECTOR): Likewise.
335 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
336 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
337 Build a new vector rather than modifying a CONST_VECTOR in-place.
338 (handle_special_swappables): Update call accordingly.
339 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
340 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
341 Build a new vector rather than modifying a CONST_VECTOR in-place.
342 (handle_special_swappables): Update call accordingly.
344 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
346 * simplify-rtx.c (simplify_const_binary_operation): Use
347 CONST_VECTOR_ELT instead of XVECEXP.
349 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
351 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
352 the selector elements to be different from the data elements
353 if the selector is a VECTOR_CST.
354 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
355 ssizetype for the selector.
357 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
359 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
360 before testing each element individually.
361 * tree-vect-generic.c (lower_vec_perm): Likewise.
363 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
365 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
366 * selftest-run-tests.c (selftest::run_tests): Call it.
367 * vector-builder.h (vector_builder::operator ==): New function.
368 (vector_builder::operator !=): Likewise.
369 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
370 (vec_perm_indices::all_from_input_p): New function.
371 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
372 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
373 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
374 instead of reading the VECTOR_CST directly. Detect whether both
375 vector inputs are the same before constructing the vec_perm_indices,
376 and update the number of inputs argument accordingly. Use the
377 utility functions added above. Only construct sel2 if we need to.
379 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
381 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
382 the broadcast of the low byte.
383 (expand_mult_highpart): Use an explicit encoding for the permutes.
384 * optabs-query.c (can_mult_highpart_p): Likewise.
385 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
386 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
387 (vectorizable_bswap): Likewise.
388 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
389 explicit encoding for the power-of-2 permutes.
390 (vect_permute_store_chain): Likewise.
391 (vect_grouped_load_supported): Likewise.
392 (vect_permute_load_chain): Likewise.
394 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
396 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
397 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
398 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
399 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
400 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
401 (vect_gen_perm_mask_any): Likewise.
403 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
405 * int-vector-builder.h: New file.
406 * vec-perm-indices.h: Include int-vector-builder.h.
407 (vec_perm_indices): Redefine as an int_vector_builder.
408 (auto_vec_perm_indices): Delete.
409 (vec_perm_builder): Redefine as a stand-alone class.
410 (vec_perm_indices::vec_perm_indices): New function.
411 (vec_perm_indices::clamp): Likewise.
412 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
413 (vec_perm_indices::new_vector): New function.
414 (vec_perm_indices::new_expanded_vector): Update for new
415 vec_perm_indices class.
416 (vec_perm_indices::rotate_inputs): New function.
417 (vec_perm_indices::all_in_range_p): Operate directly on the
418 encoded form, without computing elided elements.
419 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
420 encoding. Update for new vec_perm_indices class.
421 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
422 the given vec_perm_builder.
423 (expand_vec_perm_var): Update vec_perm_builder constructor.
424 (expand_mult_highpart): Use vec_perm_builder instead of
425 auto_vec_perm_indices.
426 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
427 vec_perm_indices instead of auto_vec_perm_indices. Use a single
428 or double series encoding as appropriate.
429 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
430 vec_perm_indices instead of auto_vec_perm_indices.
431 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
432 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
433 (vect_permute_store_chain): Likewise.
434 (vect_grouped_load_supported): Likewise.
435 (vect_permute_load_chain): Likewise.
436 (vect_shift_permute_load_chain): Likewise.
437 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
438 (vect_transform_slp_perm_load): Likewise.
439 (vect_schedule_slp_instance): Likewise.
440 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
441 (vectorizable_mask_load_store): Likewise.
442 (vectorizable_bswap): Likewise.
443 (vectorizable_store): Likewise.
444 (vectorizable_load): Likewise.
445 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
446 vec_perm_indices instead of auto_vec_perm_indices. Use
447 tree_to_vec_perm_builder to read the vector from a tree.
448 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
449 vec_perm_builder instead of a vec_perm_indices.
450 (have_whole_vector_shift): Use vec_perm_builder and
451 vec_perm_indices instead of auto_vec_perm_indices. Leave the
452 truncation to calc_vec_perm_mask_for_shift.
453 (vect_create_epilog_for_reduction): Likewise.
454 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
455 from auto_vec_perm_indices to vec_perm_indices.
456 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
457 instead of changing individual elements.
458 (aarch64_vectorize_vec_perm_const): Use new_vector to install
459 the vector in d.perm.
460 * config/arm/arm.c (expand_vec_perm_d::perm): Change
461 from auto_vec_perm_indices to vec_perm_indices.
462 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
463 instead of changing individual elements.
464 (arm_vectorize_vec_perm_const): Use new_vector to install
465 the vector in d.perm.
466 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
467 Update vec_perm_builder constructor.
468 (rs6000_expand_interleave): Likewise.
469 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
470 (rs6000_expand_interleave): Likewise.
472 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
474 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
475 to qimode could truncate the indices.
476 * optabs.c (expand_vec_perm_var): Likewise.
478 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
480 * Makefile.in (OBJS): Add vec-perm-indices.o.
481 * vec-perm-indices.h: New file.
482 * vec-perm-indices.c: Likewise.
483 * target.h (vec_perm_indices): Replace with a forward class
485 (auto_vec_perm_indices): Move to vec-perm-indices.h.
486 * optabs.h: Include vec-perm-indices.h.
487 (expand_vec_perm): Delete.
488 (selector_fits_mode_p, expand_vec_perm_var): Declare.
489 (expand_vec_perm_const): Declare.
490 * target.def (vec_perm_const_ok): Replace with...
491 (vec_perm_const): ...this new hook.
492 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
493 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
494 * doc/tm.texi: Regenerate.
495 * optabs.def (vec_perm_const): Delete.
496 * doc/md.texi (vec_perm_const): Likewise.
497 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
498 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
499 expand_vec_perm for constant permutation vectors. Assert that
500 the mode of variable permutation vectors is the integer equivalent
501 of the mode that is being permuted.
502 * optabs-query.h (selector_fits_mode_p): Declare.
503 * optabs-query.c: Include vec-perm-indices.h.
504 (selector_fits_mode_p): New function.
505 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
506 is defined, instead of checking whether the vec_perm_const_optab
507 exists. Use targetm.vectorize.vec_perm_const instead of
508 targetm.vectorize.vec_perm_const_ok. Check whether the indices
509 fit in the vector mode before using a variable permute.
510 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
511 vec_perm_indices instead of an rtx.
512 (expand_vec_perm): Replace with...
513 (expand_vec_perm_const): ...this new function. Take the selector
514 as a vec_perm_indices rather than an rtx. Also take the mode of
515 the selector. Update call to shift_amt_for_vec_perm_mask.
516 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
517 Use vec_perm_indices::new_expanded_vector to expand the original
518 selector into bytes. Check whether the indices fit in the vector
519 mode before using a variable permute.
520 (expand_vec_perm_var): Make global.
521 (expand_mult_highpart): Use expand_vec_perm_const.
522 * fold-const.c: Includes vec-perm-indices.h.
523 * tree-ssa-forwprop.c: Likewise.
524 * tree-vect-data-refs.c: Likewise.
525 * tree-vect-generic.c: Likewise.
526 * tree-vect-loop.c: Likewise.
527 * tree-vect-slp.c: Likewise.
528 * tree-vect-stmts.c: Likewise.
529 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
531 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
532 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
533 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
534 (aarch64_vectorize_vec_perm_const): ...this new function.
535 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
536 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
537 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
538 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
539 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
540 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
541 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
543 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
544 check for NEON modes.
545 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
546 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
547 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
548 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
550 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
551 the old VEC_PERM_CONST conditions.
552 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
553 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
554 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
555 (ia64_vectorize_vec_perm_const_ok): Merge into...
556 (ia64_vectorize_vec_perm_const): ...this new function.
557 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
558 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
559 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
560 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
561 * config/mips/mips.c (mips_expand_vec_perm_const)
562 (mips_vectorize_vec_perm_const_ok): Merge into...
563 (mips_vectorize_vec_perm_const): ...this new function.
564 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
565 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
566 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
567 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
568 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
569 (rs6000_expand_vec_perm_const): Delete.
570 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
572 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
573 (altivec_expand_vec_perm_const_le): Take each operand individually.
574 Operate on constant selectors rather than rtxes.
575 (altivec_expand_vec_perm_const): Likewise. Update call to
576 altivec_expand_vec_perm_const_le.
577 (rs6000_expand_vec_perm_const): Delete.
578 (rs6000_vectorize_vec_perm_const_ok): Delete.
579 (rs6000_vectorize_vec_perm_const): New function.
580 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
581 an element count and rtx array.
582 (rs6000_expand_extract_even): Update call accordingly.
583 (rs6000_expand_interleave): Likewise.
584 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
585 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
586 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
587 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
588 (rs6000_expand_vec_perm_const): Delete.
589 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
590 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
591 (altivec_expand_vec_perm_const_le): Take each operand individually.
592 Operate on constant selectors rather than rtxes.
593 (altivec_expand_vec_perm_const): Likewise. Update call to
594 altivec_expand_vec_perm_const_le.
595 (rs6000_expand_vec_perm_const): Delete.
596 (rs6000_vectorize_vec_perm_const_ok): Delete.
597 (rs6000_vectorize_vec_perm_const): New function. Remove stray
598 reference to the SPE evmerge intructions.
599 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
600 an element count and rtx array.
601 (rs6000_expand_extract_even): Update call accordingly.
602 (rs6000_expand_interleave): Likewise.
603 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
604 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
606 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
608 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
610 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
611 vector mode and that that mode matches the mode of the data
613 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
614 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
615 directly using expand_vec_perm_1 when forcing selectors into
617 (expand_vec_perm_var): New function, split out from expand_vec_perm.
619 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
621 * optabs-query.h (can_vec_perm_p): Delete.
622 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
623 * optabs-query.c (can_vec_perm_p): Split into...
624 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
625 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
626 particular selector is valid.
627 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
628 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
629 (vect_grouped_load_supported): Likewise.
630 (vect_shift_permute_load_chain): Likewise.
631 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
632 (vect_transform_slp_perm_load): Likewise.
633 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
634 (vectorizable_bswap): Likewise.
635 (vect_gen_perm_mask_checked): Likewise.
636 * fold-const.c (fold_ternary_loc): Likewise. Don't take
637 implementations of variable permutation vectors into account
638 when deciding which selector to use.
639 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
640 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
641 with a false third argument.
642 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
643 to test whether the constant selector is valid and can_vec_perm_var_p
644 to test whether a variable selector is valid.
646 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
648 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
649 * optabs-query.c (can_vec_perm_p): Likewise.
650 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
651 instead of vec_perm_indices.
652 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
653 (vect_gen_perm_mask_checked): Likewise,
654 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
655 (vect_gen_perm_mask_checked): Likewise,
657 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
659 * optabs-query.h (qimode_for_vec_perm): Declare.
660 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
661 (qimode_for_vec_perm): ...this new function.
662 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
664 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
666 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
667 does not have a conditional at the top.
669 2018-01-02 Richard Biener <rguenther@suse.de>
671 * ipa-inline.c (big_speedup_p): Fix expression.
673 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
676 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
679 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
683 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
684 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
685 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
686 cond_taken_branch_cost 3->4.
688 2018-01-01 Jakub Jelinek <jakub@redhat.com>
690 PR tree-optimization/83581
691 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
692 TODO_cleanup_cfg if any changes have been made.
695 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
696 convert_modes if target mode has the right side, but different mode
700 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
701 last argument when extracting from CONCAT. If either from_real or
702 from_imag is NULL, use expansion through memory. If result is not
703 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
704 the parts directly to inner mode, if even that fails, use expansion
708 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
709 check for bswap in mode rather than HImode and use that in expand_unop
712 2017-12-31 Jakub Jelinek <jakub@redhat.com>
715 * config/i386/i386.c (ix86_attribute_table): Remove excess
716 initializer for "shared" attribute.
718 2017-12-30 Tom de Vries <tom@codesourcery.com>
721 * omp-expand.c (expand_omp_target): If in_lto_p, mark offload_funcs with
723 * lto-streamer-out.c (prune_offload_funcs): New function. Remove
724 offload_funcs entries that no longer have a corresponding cgraph_node.
725 Mark the remaining ones as DECL_PRESERVE_P.
726 (output_lto): Call prune_offload_funcs.
728 2017-12-30 Jakub Jelinek <jakub@redhat.com>
730 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
731 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>,
732 vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
733 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
734 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
735 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1, vpdpbusd_<mode>,
736 vpdpbusd_<mode>_mask, vpdpbusd_<mode>_maskz, vpdpbusd_<mode>_maskz_1,
737 vpdpbusds_<mode>, vpdpbusds_<mode>_mask, vpdpbusds_<mode>_maskz,
738 vpdpbusds_<mode>_maskz_1, vpdpwssd_<mode>, vpdpwssd_<mode>_mask,
739 vpdpwssd_<mode>_maskz, vpdpwssd_<mode>_maskz_1, vpdpwssds_<mode>,
740 vpdpwssds_<mode>_mask, vpdpwssds_<mode>_maskz,
741 vpdpwssds_<mode>_maskz_1, vaesdec_<mode>, vaesdeclast_<mode>,
742 vaesenc_<mode>, vpclmulqdq_<mode>,
743 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Formatting fixes.
745 2017-12-28 Michael Meissner <meissner@linux.vnet.ibm.com>
747 * builtins.def: (_Float<N> and _Float<N>X BUILT_IN_CEIL): Add
748 _Float<N> and _Float<N>X variants for rounding built-in
750 (_Float<N> and _Float<N>X BUILT_IN_FLOOR): Likewise.
751 (_Float<N> and _Float<N>X BUILT_IN_NEARBYINT): Likewise.
752 (_Float<N> and _Float<N>X BUILT_IN_RINT): Likewise.
753 (_Float<N> and _Float<N>X BUILT_IN_ROUND): Likewise.
754 (_Float<N> and _Float<N>X BUILT_IN_TRUNC): Likewise.
755 * builtins.c (mathfn_built_in_2): Likewise.
756 * internal-fn.def (CEIL): Likewise.
758 (NEARBYINT): Likewise.
762 * convert.c (convert_to_integer_1): Likewise.
763 * fold-const.c (tree_call_nonnegative_warnv_p): Likewise.
764 (integer_valued_real_call_p): Likewise.
765 * fold-const-call.c (fold_const_call_ss): Likewise.
766 * gencfn-macros.c (print_case_cfn): Change CFN and operator
767 printers to take a const char * suffix instead of a bool.
768 (print_define_operator_list): Likewise.
769 (fltall_suffixes): New list of suffixes, that include the
770 traditional suffixes as well as all of the _Float<N> and
772 (main): For _Float<N> and _Float<N>X functions, emit both
773 <name>_FN and <name>_ALL variants. The <macro>_FN variant only
774 has the _Float<N> and _Float<N>X case names or operators. The
775 <name>_ALL variant has both the traditional and the
776 _Float<N>/_Float<N>X case names or operators.
777 * match.pd (COPYSIGN optimizations): Provide optimizations for
778 _Float<N> and _Float<N>X types where possible.
779 (MIN/MAX optimizations): Likewise.
780 (sqrt optimizations): Likewise.
781 (rounding optimizations): Likewise.
783 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
785 * config/powerpcspe/powerpcspe.md (ieee_128bit_negative_zero): Use
786 gen_int_mode rather than GEN_INT.
787 * config/rs6000/rs6000.md (ieee_128bit_negative_zero): Likewise.
789 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
791 * emit-rtl.c (gen_const_vec_series): Use valid_for_const_vector_p
792 instead of CONSTANT_P.
793 (gen_vec_series): Likewise.
794 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
796 2017-12-28 Andreas Schwab <schwab@linux-m68k.org>
798 * config/m68k/m68k.md (ashrdi3_const1, lshrdi3_const1): Add
801 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
803 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
804 (aarch64_ld1x2<VDC:mode>): Likewise.
805 (aarch64_simd_ld1<mode>_x2): Likewise.
806 (aarch64_simd_ld1<mode>_x2): Likewise.
807 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
808 (vld1_s8_x2): Likewise.
809 (vld1_u16_x2): Likewise.
810 (vld1_s16_x2): Likewise.
811 (vld1_u32_x2): Likewise.
812 (vld1_s32_x2): Likewise.
813 (vld1_u64_x2): Likewise.
814 (vld1_s64_x2): Likewise.
815 (vld1_f16_x2): Likewise.
816 (vld1_f32_x2): Likewise.
817 (vld1_f64_x2): Likewise.
818 (vld1_p8_x2): Likewise.
819 (vld1_p16_x2): Likewise.
820 (vld1_p64_x2): Likewise.
821 (vld1q_u8_x2): Likewise.
822 (vld1q_s8_x2): Likewise.
823 (vld1q_u16_x2): Likewise.
824 (vld1q_s16_x2): Likewise.
825 (vld1q_u32_x2): Likewise.
826 (vld1q_s32_x2): Likewise.
827 (vld1q_u64_x2): Likewise.
828 (vld1q_s64_x2): Likewise.
829 (vld1q_f16_x2): Likewise.
830 (vld1q_f32_x2): Likewise.
831 (vld1q_f64_x2): Likewise.
832 (vld1q_p8_x2): Likewise.
833 (vld1q_p16_x2): Likewise.
834 (vld1q_p64_x2): Likewise.
836 2017-12-27 Martin Liska <mliska@suse.cz>
838 PR tree-optimization/83552
839 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
840 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
842 2017-12-27 Tom de Vries <tom@codesourcery.com>
844 * config/nvptx/nvptx.c (nvptx_option_override): Disable
845 -gstatement-frontiers.
847 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
849 PR rtl-optimization/83513
850 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
851 before priority comparison.
853 2017-12-25 Jakub Jelinek <jakub@redhat.com>
856 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
857 ix86_isa_flags2 to ix86_isa_flags.
858 * config/i386/i386-c.c (ix86_target_macros_internal): Test
859 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
860 isa_flags rather than isa_flags2.
861 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
862 and -mavx512bitalg from isa2_opts to isa_opts.
863 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
864 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
865 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
867 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
868 OPTION_MASK_ISA_AVX512F ored with another option similarly to
869 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
870 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
871 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
872 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
873 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
874 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
875 builtins from bdesc_args2 to bdesc_args section.
876 (__builtin_ia32_compressstoreuqi512_mask,
877 __builtin_ia32_compressstoreuhi512_mask,
878 __builtin_ia32_compressstoreuqi256_mask,
879 __builtin_ia32_expandloadqi512_mask,
880 __builtin_ia32_expandloadqi512_maskz,
881 __builtin_ia32_expandloadhi512_mask,
882 __builtin_ia32_expandloadhi512_maskz,
883 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
884 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
885 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
886 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
887 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
888 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
889 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
890 __builtin_ia32_vpshldv_v32hi_maskz,
891 __builtin_ia32_vpopcountb_v64qi_mask,
892 __builtin_ia32_vpopcountw_v32hi_mask,
893 __builtin_ia32_vpshufbitqmb512_mask,
894 __builtin_ia32_vpshufbitqmb256_mask): Add
895 " | OPTION_MASK_ISA_AVX512BW".
896 (__builtin_ia32_expandloadqi256_mask,
897 __builtin_ia32_expandloadqi256_maskz,
898 __builtin_ia32_vpopcountb_v32qi_mask): Add
899 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
900 (__builtin_ia32_expandloadhi256_mask,
901 __builtin_ia32_expandloadhi256_maskz,
902 __builtin_ia32_expandloadqi128_mask,
903 __builtin_ia32_expandloadqi128_maskz,
904 __builtin_ia32_expandloadhi128_mask,
905 __builtin_ia32_expandloadhi128_maskz,
906 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
907 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
908 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
909 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
910 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
911 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
912 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
913 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
914 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
915 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
916 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
917 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
918 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
919 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
920 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
921 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
922 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
923 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
924 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
925 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
926 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
927 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
928 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
929 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
930 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
931 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
932 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
933 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
934 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
935 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
936 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
937 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
938 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
939 __builtin_ia32_vpopcountw_v8hi_mask): Add
940 " | OPTION_MASK_ISA_AVX512VL".
941 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
942 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
943 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
944 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
945 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
946 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
947 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
948 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
949 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
950 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
951 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
952 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
953 _mm_mask_bitshuffle_epi64_mask): Likewise.
954 * common/config/i386/i386-common.c
955 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
956 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
957 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
958 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
959 OPTION_MASK_ISA_AVX512BITALG_UNSET.
960 (OPTION_MASK_ISA2_AVX512F_UNSET,
961 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
962 (ix86_handle_option): For -mno-general-regs-only, clear from
963 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
964 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
965 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
966 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
967 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
968 for moving from ix86_isa_flags2 to ix86_isa_flags.
970 2017-12-23 Jakub Jelinek <jakub@redhat.com>
973 * fold-const.c (struct contains_label_data): New type.
974 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
975 inside of a SWITCH_BODY seen during the walk.
976 (contains_label_p): Use walk_tree instead of
977 walk_tree_without_duplicates, prepare data for contains_label_1 and
980 2017-12-22 Jakub Jelinek <jakub@redhat.com>
983 * tree-iterator.c (alloc_stmt_list): Start with cleared
984 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
988 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
989 enable avx512f explicitly in #pragma GCC target.
990 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
991 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
992 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
993 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
994 __builtin_ia32_vpdpbusds_v8si_mask,
995 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
996 __builtin_ia32_vpdpbusds_v4si_mask,
997 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
998 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
999 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
1000 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
1001 __builtin_ia32_vpdpwssds_v8si_mask,
1002 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
1003 __builtin_ia32_vpdpwssds_v4si_mask,
1004 __builtin_ia32_vpdpwssds_v4si_maskz): Use
1005 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
1006 just OPTION_MASK_ISA_AVX512VNNI.
1008 2017-12-22 Martin Jambor <mjambor@suse.cz>
1011 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
1014 2017-12-22 Julia Koval <julia.koval@intel.com>
1015 Sebastian Peryt <sebastian.peryt@intel.com>
1017 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
1018 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
1019 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
1020 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
1021 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
1022 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
1023 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
1024 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
1025 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
1026 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
1027 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
1028 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
1029 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
1030 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
1031 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
1032 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
1033 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
1034 _mm256_maskz_popcnt_epi64): New intrinsics.
1035 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
1036 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
1037 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
1038 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
1039 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
1040 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
1041 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
1042 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
1043 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
1044 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
1045 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
1046 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
1047 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
1048 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
1049 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
1050 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
1051 __builtin_ia32_vpshufbitqmb256_mask,
1052 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
1053 * config/i386/i386-c.c (__AVX512BITALG__): New.
1054 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
1055 (ix86_valid_target_attribute_inner_p): Ditto.
1056 (ix86_expand_args_builtin): Handle new types.
1057 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
1058 * config/i386/i386.opt: Add -mavx512bitalg.
1059 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
1060 avx512bitalgintrin.h.
1061 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
1062 (vpopcount<mode><mask_name>): Add more types.
1063 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
1064 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
1066 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1068 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
1069 Or in OPTION_MASK_ISA_AVX512F_SET.
1070 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1071 OPTION_MASK_ISA_AVX512VNNI_UNSET.
1072 (ix86_handle_option): Adjust for
1073 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
1074 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
1076 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
1077 isa_flag instead of isa_flag2.
1078 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
1079 isa_opts2 to isa_opts.
1080 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
1083 2017-12-22 Mike Stump <mikestump@comcast.net>
1084 Eric Botcazou <ebotcazou@adacore.com>
1086 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
1088 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1091 * config/i386/i386.c (ix86_function_arg_boundary): Return
1092 PARM_BOUNDARY for TYPE_EMPTY_P types.
1095 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
1096 if navail is >= dir.len.
1098 2017-12-21 Steve Ellcey <sellcey@cavium.com>
1100 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
1103 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1105 PR rtl-optimization/80747
1106 PR rtl-optimization/83512
1107 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
1108 succ edge from ENTRY, copy partition from e->dest to the newly
1110 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
1111 ENTRY, use BB_PARTITION of its successor block as current_partition.
1112 Don't copy partition when splitting succ edge from ENTRY.
1114 PR tree-optimization/83523
1115 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
1116 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
1117 (convert_mult_to_fma): Likewise.
1119 PR tree-optimization/83521
1120 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
1121 gimple_build_assign without code on result of
1122 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
1123 a VIEW_CONVERT_EXPR.
1125 2017-12-21 Andrew Pinski <apinski@cavium.com>
1126 Steve Ellcey <sellcey@cavium.com>
1128 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
1129 multi-arch for ilp32.
1131 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
1134 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
1135 constraints to operand 2.
1136 (*ashl<mode>3_mask_1): Ditto.
1137 (*<shift_insn><mode>3_mask): Ditto.
1138 (*<shift_insn><mode>3_mask_1): Ditto.
1139 (*<rotate_insn><mode>3_mask): Ditto.
1140 (*<rotate_insn><mode>3_mask_1): Ditto.
1142 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
1144 * reorg.c (make_return_insns): Reemit each insn with its own location.
1146 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1148 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
1149 generation for cases where splatting a value is not useful.
1150 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1151 across a vec_duplicate and a paradoxical subreg forming a vector
1152 mode to a vec_concat.
1154 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1156 * combine.c (simplify_set): Do not transform subregs to zero_extends
1157 if the destination is not a scalar int mode.
1159 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
1162 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
1163 shared zero if the input has overflowed.
1165 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1167 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
1168 support for Cortex-A55 and Cortex-A75.
1170 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1172 * common/config/arm/arm-common.c (compare_opt_names): Add function
1173 comment. Use strcmp instead of manual loop.
1175 2017-12-21 Martin Liska <mliska@suse.cz>
1177 PR gcov-profile/83509
1178 * gcov-dump.c (dump_gcov_file): Do not read info about
1179 support_unexecuted_blocks for gcda files.
1181 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1183 PR rtl-optimization/82973
1184 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
1185 (valid_for_const_vector_p): ... this.
1186 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
1187 (valid_for_const_vector_p): ... this. Adjust function comment.
1188 (gen_vec_duplicate): Adjust caller.
1189 * optabs.c (expand_vector_broadcast): Likewise.
1190 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
1191 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
1193 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
1194 instead of GET_CODE == CONST_FIXED.
1195 (simplify_subreg): Use CONST_FIXED_P macro instead of
1196 GET_CODE == CONST_FIXED.
1199 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
1200 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
1201 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
1202 (ix86_option_override_internal): Adjust for
1203 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
1204 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
1205 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
1206 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
1207 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
1208 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
1209 def_builtin2 for CET builtins.
1210 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
1211 up formatting in IX86_BUILTIN_RDPID code.
1212 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
1213 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
1214 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
1215 ix86_isa_flags2 to ix86_isa_flags.
1216 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
1218 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
1219 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
1220 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
1222 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
1223 Or in OPTION_MASK_ISA_AVX512F_SET.
1224 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1225 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
1226 (ix86_handle_option): Adjust for
1227 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
1228 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
1230 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1231 Alan Hayward <alan.hayward@arm.com>
1232 David Sherwood <david.sherwood@arm.com>
1234 * tree-data-ref.h (prune_runtime_alias_test_list): Take the
1235 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1236 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
1237 Track polynomial offsets.
1239 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1240 Alan Hayward <alan.hayward@arm.com>
1241 David Sherwood <david.sherwood@arm.com>
1243 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
1244 Treat drb->init as a poly_int. Fail if its misalignment wrt
1245 vector_alignment isn't known.
1247 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1248 Alan Hayward <alan.hayward@arm.com>
1249 David Sherwood <david.sherwood@arm.com>
1251 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
1253 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
1254 (vect_loop_versioning): Take the loop versioning threshold as a
1256 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
1257 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
1258 versioning_threshold.
1259 (vect_analyze_loop_2): Compute the loop versioning threshold
1260 whenever loop versioning is needed, and store it in the new
1261 field rather than combining it with the cost model threshold.
1262 (vect_transform_loop): Update call to vect_loop_versioning.
1263 Try to combine the loop versioning and cost thresholds here.
1265 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1266 Alan Hayward <alan.hayward@arm.com>
1267 David Sherwood <david.sherwood@arm.com>
1269 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
1270 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
1271 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
1272 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
1273 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
1274 (group_compare_offset): Update accordingly.
1275 (split_small_address_groups_p): Likewise.
1276 (record_use): Take addr_offset as a poly_uint64 rather than
1277 an unsigned HOST_WIDE_INT.
1278 (strip_offset): Return the offset as a poly_uint64 rather than
1279 an unsigned HOST_WIDE_INT.
1280 (record_group_use, split_address_groups): Track polynomial offsets.
1281 (add_iv_candidate_for_use): Likewise.
1282 (addr_offset_valid_p): Take the offset as a poly_int64 rather
1283 than a HOST_WIDE_INT.
1284 (strip_offset_1): Return the offset as a poly_int64 rather than
1287 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1288 Alan Hayward <alan.hayward@arm.com>
1289 David Sherwood <david.sherwood@arm.com>
1291 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
1292 rather than a HOST_WIDE_INT.
1293 * tree.c (get_binfo_at_offset): Likewise.
1295 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1296 Alan Hayward <alan.hayward@arm.com>
1297 David Sherwood <david.sherwood@arm.com>
1299 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
1300 rather than a HOST_WIDE_INT.
1301 * tree-sra.c (build_ref_for_offset): Likewise.
1303 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1304 Alan Hayward <alan.hayward@arm.com>
1305 David Sherwood <david.sherwood@arm.com>
1307 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
1309 * tree.c (mem_ref_offset): Likewise.
1310 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
1311 * builtins.c (get_object_alignment_2): Likewise.
1312 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
1313 * gimple-fold.c (get_base_constructor): Likewise.
1314 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
1315 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1317 * ipa-polymorphic-call.c
1318 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1319 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
1320 (get_ancestor_addr_info): Likewise.
1321 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
1322 * match.pd: Likewise.
1323 * tree-data-ref.c (dr_analyze_innermost): Likewise.
1324 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1325 * tree-eh.c (tree_could_trap_p): Likewise.
1326 * tree-object-size.c (addr_object_size): Likewise.
1327 * tree-ssa-address.c (copy_ref_info): Likewise.
1328 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
1329 (indirect_refs_may_alias_p): Likewise.
1330 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1331 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
1332 (non_rewritable_mem_ref_base): Likewise.
1333 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1334 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1335 * varasm.c (decode_addr_const): 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 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
1342 offsets for MEM_REFs.
1344 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1345 Alan Hayward <alan.hayward@arm.com>
1346 David Sherwood <david.sherwood@arm.com>
1348 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
1349 as a poly_uint64 rather than an unsigned int.
1350 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
1352 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1353 Alan Hayward <alan.hayward@arm.com>
1354 David Sherwood <david.sherwood@arm.com>
1356 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
1358 (decode_addr_const): Update accordingly.
1360 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1361 Alan Hayward <alan.hayward@arm.com>
1362 David Sherwood <david.sherwood@arm.com>
1364 * tree.h (bit_field_size, bit_field_offset): New functions.
1365 * hsa-gen.c (gen_hsa_addr): Use them.
1366 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
1367 (simplify_vector_constructor): Likewise.
1368 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1369 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
1370 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
1371 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
1374 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1375 Alan Hayward <alan.hayward@arm.com>
1376 David Sherwood <david.sherwood@arm.com>
1378 * expr.h (emit_group_load, emit_group_load_into_temps)
1379 (emit_group_store): Take the size as a poly_int64 rather than an int.
1380 * expr.c (emit_group_load_1, emit_group_load): Likewise.
1381 (emit_group_load_into_temp, emit_group_store): Likewise.
1383 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1384 Alan Hayward <alan.hayward@arm.com>
1385 David Sherwood <david.sherwood@arm.com>
1387 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
1388 an unsigned int to a poly_uint64.
1389 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
1390 sizes as poly_uint64s rather than unsigned ints.
1391 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
1394 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1395 Alan Hayward <alan.hayward@arm.com>
1396 David Sherwood <david.sherwood@arm.com>
1398 * lra-constraints.c (emit_inc): Change inc_amount from an 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 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
1407 (add_stack_var, stack_var_cmp, partition_stack_vars)
1408 (dump_stack_var_partition): Update accordingly.
1409 (alloc_stack_frame_space): Take the size as a poly_int64 rather
1410 than a HOST_WIDE_INT.
1411 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
1412 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
1413 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
1414 rather than a HOST_WIDE_INT.
1416 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1417 Alan Hayward <alan.hayward@arm.com>
1418 David Sherwood <david.sherwood@arm.com>
1420 * target.def (return_pops_args): Treat both the input and output
1421 sizes as poly_int64s rather than HOST_WIDE_INTS.
1422 * targhooks.h (default_return_pops_args): Update accordingly.
1423 * targhooks.c (default_return_pops_args): Likewise.
1424 * doc/tm.texi: Regenerate.
1425 * emit-rtl.h (incoming_args): Change pops_args, size and
1426 outgoing_args_size from int to poly_int64_pod.
1427 * function.h (expr_status): Change x_pending_stack_adjust and
1428 x_stack_pointer_delta from int to poly_int64.
1429 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
1430 (ARGS_SIZE_RTX): Update accordingly.
1431 * calls.c (highest_outgoing_arg_in_use): Change from int to
1433 (stack_usage_watermark, stored_args_watermark): New variables.
1434 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
1435 (emit_call_1): Change the stack_size and rounded_stack_size
1436 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
1438 (save_fixed_argument_area): Check stack_usage_watermark.
1439 (initialize_argument_information): Change old_pending_adj from
1440 a HOST_WIDE_INT * to a poly_int64_pod *.
1441 (compute_argument_block_size): Return the size as a poly_int64
1443 (finalize_must_preallocate): Track polynomial argument sizes.
1444 (compute_argument_addresses): Likewise.
1445 (internal_arg_pointer_based_exp): Track polynomial offsets.
1446 (mem_overlaps_already_clobbered_arg_p): Rename to...
1447 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
1448 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1449 Check stored_args_used_watermark.
1450 (load_register_parameters): Update accordingly.
1451 (check_sibcall_argument_overlap_1): Likewise.
1452 (combine_pending_stack_adjustment_and_call): Take the unadjusted
1453 args size as a poly_int64 rather than an int. Return a bool
1454 indicating whether the optimization was possible and return
1455 the new adjustment by reference.
1456 (check_sibcall_argument_overlap): Track polynomail argument sizes.
1457 Update stored_args_watermark.
1458 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
1459 (expand_call): Likewise. Maintain stack_usage_watermark and
1460 stored_args_watermark. Update calls to
1461 combine_pending_stack_adjustment_and_call.
1462 (emit_library_call_value_1): Handle polynomial argument sizes.
1463 Call stack_region_maybe_used_p and mark_stack_region_used.
1464 Maintain stack_usage_watermark.
1465 (store_one_arg): Likewise. Update call to
1466 mem_overlaps_already_clobbered_arg_p.
1467 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
1469 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
1470 * config/microblaze/microblaze.c (microblaze_function_prologue):
1472 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
1473 TARGET_RETURN_POPS_ARGS interface.
1474 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
1476 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
1477 * config/i386/i386.c (ix86_return_pops_args): Update for new
1478 TARGET_RETURN_POPS_ARGS interface.
1479 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
1480 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
1481 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
1482 TARGET_RETURN_POPS_ARGS interface.
1483 * config/vax/vax.c (vax_return_pops_args): Likewise.
1484 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
1485 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
1486 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
1487 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1488 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1489 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
1490 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1491 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1492 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
1493 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
1494 and x_stack_pointer_delta from int to poly_int64.
1495 * dojump.c (do_pending_stack_adjust): Update accordingly.
1496 * explow.c (allocate_dynamic_stack_space): Handle polynomial
1497 stack_pointer_deltas.
1498 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
1499 (pad_to_arg_alignment): Track polynomial offsets.
1500 (assign_parm_find_stack_rtl): Likewise.
1501 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
1502 * toplev.c (output_stack_usage): Update reference to
1503 current_function_pushed_stack_size.
1505 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1506 Alan Hayward <alan.hayward@arm.com>
1507 David Sherwood <david.sherwood@arm.com>
1509 * function.c (in_arg_offset, var_offset, dynamic_offset)
1510 (out_arg_offset, cfa_offset): Change from int to poly_int64.
1511 (instantiate_new_reg): Return the new offset as a poly_int64_pod
1512 rather than a HOST_WIDE_INT.
1513 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
1514 (instantiate_virtual_regs_in_insn): Likewise.
1516 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1517 Alan Hayward <alan.hayward@arm.com>
1518 David Sherwood <david.sherwood@arm.com>
1520 * rtl.h (get_args_size, add_args_size_note): New functions.
1521 (find_args_size_adjust): Return a poly_int64 rather than a
1523 (fixup_args_size_notes): Likewise. Make the same change to the
1524 end_args_size parameter.
1525 * rtlanal.c (get_args_size, add_args_size_note): New functions.
1526 * builtins.c (expand_builtin_trap): Use add_args_size_note.
1527 * calls.c (emit_call_1): Likewise.
1528 * explow.c (adjust_stack_1): Likewise.
1529 * cfgcleanup.c (old_insns_match_p): Update use of
1530 find_args_size_adjust.
1531 * combine.c (distribute_notes): Track polynomial arg sizes.
1532 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
1533 end_true_args_size, beg_delay_args_size and end_delay_args_size
1534 from HOST_WIDE_INT to poly_int64.
1535 (add_cfi_args_size): Take the args_size as a poly_int64 rather
1536 than a HOST_WIDE_INT.
1537 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
1538 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
1539 polynomial arg sizes.
1540 * emit-rtl.c (try_split): Use get_args_size.
1541 * recog.c (peep2_attempt): Likewise.
1542 * reload1.c (reload_as_needed): Likewise.
1543 * expr.c (find_args_size_adjust): Return the adjustment as a
1544 poly_int64 rather than a HOST_WIDE_INT.
1545 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
1546 to a poly_int64 and change the return type in the same way.
1547 (emit_single_push_insn): Track polynomial arg sizes.
1549 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1550 Alan Hayward <alan.hayward@arm.com>
1551 David Sherwood <david.sherwood@arm.com>
1553 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
1554 from HOST_WIDE_INT to poly_int64.
1555 * expr.c (push_block, emit_push_insn): Likewise.
1557 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1558 Alan Hayward <alan.hayward@arm.com>
1559 David Sherwood <david.sherwood@arm.com>
1561 * function.h (frame_space): Change start and length from HOST_WIDE_INT
1563 (get_frame_size): Return the size as a poly_int64 rather than a
1565 (frame_offset_overflow): Take the offset as a poly_int64 rather
1566 than a HOST_WIDE_INT.
1567 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1568 (assign_stack_temp): Likewise for the size.
1569 * function.c (get_frame_size): Return a poly_int64 rather than
1571 (frame_offset_overflow): Take the offset as a poly_int64 rather
1572 than a HOST_WIDE_INT.
1573 (try_fit_stack_local): Take the start, length and size as poly_int64s
1574 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
1575 rather than a HOST_WIDE_INT.
1576 (add_frame_space): Take the start and end as poly_int64s rather than
1578 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1579 (assign_stack_temp): Likewise for the size.
1580 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
1582 (find_temp_slot_from_address): Handle polynomial offsets.
1583 (combine_temp_slots): Likewise.
1584 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
1586 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
1587 poly_int64 rather than a HOST_WIDE_INT.
1588 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
1589 than a HOST_WIDE_INT.
1590 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
1591 polynomial frame offsets.
1592 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
1593 as a poly_int64 rather than an int.
1594 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
1595 * config/v850/v850-protos.h (compute_frame_size): Likewise.
1596 * config/v850/v850.c (compute_frame_size): Likewise.
1597 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
1598 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
1599 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
1600 * config/pa/pa.c (pa_compute_frame_size): Likewise.
1601 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
1602 rather than a HOST_WIDE_INT.
1603 * explow.c (get_dynamic_stack_base): Likewise.
1604 * final.c (final_start_function): Use the constant lower bound
1605 of the frame size for -Wframe-larger-than.
1606 * ira.c (do_reload): Adjust for new get_frame_size return type.
1607 * lra.c (lra): Likewise.
1608 * reload1.c (reload): Likewise.
1609 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
1610 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
1611 * rtlanal.c (get_initial_register_offset): Return the offset as
1612 a poly_int64 rather than a HOST_WIDE_INT.
1614 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1615 Alan Hayward <alan.hayward@arm.com>
1616 David Sherwood <david.sherwood@arm.com>
1618 * reload1.c (elim_table): Change initial_offset, offset and
1619 previous_offset from HOST_WIDE_INT to poly_int64_pod.
1620 (offsets_at): Change the target array's element type from
1621 HOST_WIDE_INT to poly_int64_pod.
1622 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
1623 (elimination_costs_in_insn, update_eliminable_offsets)
1624 (verify_initial_elim_offsets, set_offsets_for_label)
1625 (init_eliminable_invariants): Update after above changes.
1627 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1628 Alan Hayward <alan.hayward@arm.com>
1629 David Sherwood <david.sherwood@arm.com>
1631 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
1632 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
1633 (decomposition): Change start and end from HOST_WIDE_INT
1635 (decompose, immune_p): Update accordingly.
1636 (find_inc_amount): Return a poly_int64 rather than an int.
1637 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
1640 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1641 Alan Hayward <alan.hayward@arm.com>
1642 David Sherwood <david.sherwood@arm.com>
1644 * tree.h (get_inner_reference): Return the bitsize and bitpos
1645 as poly_int64_pods rather than HOST_WIDE_INT.
1646 * fold-const.h (ptr_difference_const): Return the pointer difference
1647 as a poly_int64_pod rather than a HOST_WIDE_INT.
1648 * expr.c (get_inner_reference): Return the bitsize and bitpos
1649 as poly_int64_pods rather than HOST_WIDE_INT.
1650 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
1652 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
1653 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
1654 (optimize_bit_field_compare): Update call to get_inner_reference.
1655 (decode_field_reference): Likewise.
1656 (fold_unary_loc): Track polynomial offsets and sizes.
1657 (split_address_to_core_and_offset): Return the bitpos as a
1658 poly_int64_pod rather than a HOST_WIDE_INT.
1659 (ptr_difference_const): Likewise for the pointer difference.
1660 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
1661 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
1662 * dbxout.c (dbxout_expand_expr): Likewise.
1663 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
1664 (loc_list_from_tree_1, fortran_common): Likewise.
1665 * gimple-laddress.c (pass_laddress::execute): Likewise.
1666 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
1667 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
1668 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1669 * tree-affine.c (tree_to_aff_combination): Likewise.
1670 (get_inner_reference_aff): Likewise.
1671 * tree-data-ref.c (split_constant_offset_1): Likewise.
1672 (dr_analyze_innermost): Likewise.
1673 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
1674 * tree-sra.c (ipa_sra_check_caller): Likewise.
1675 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1676 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
1677 (instrument_bool_enum_load, instrument_object_size): Likewise.
1678 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
1679 to get_inner_reference.
1680 * hsa-gen.c (gen_hsa_addr): Likewise.
1681 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
1682 * tsan.c (instrument_expr): Likewise.
1683 * match.pd: Update call to ptr_difference_const.
1685 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1686 Alan Hayward <alan.hayward@arm.com>
1687 David Sherwood <david.sherwood@arm.com>
1689 * fold-const.c (fold_comparison): Track sizes and offsets as
1690 poly_int64s rather than HOST_WIDE_INTs when folding address
1693 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1694 Alan Hayward <alan.hayward@arm.com>
1695 David Sherwood <david.sherwood@arm.com>
1697 * expr.h (get_bit_range): Return the bitstart and bitend as
1698 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
1699 as a poly_int64 rather than a HOST_WIDE_INT.
1700 * expr.c (get_bit_range): Likewise.
1701 (expand_assignment): Update call accordingly.
1702 * fold-const.c (optimize_bit_field_compare): Likewise.
1704 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1705 Alan Hayward <alan.hayward@arm.com>
1706 David Sherwood <david.sherwood@arm.com>
1708 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
1709 (aarch64_legitimate_address_p): Use it instead of an rtx code,
1710 as an optional final parameter.
1711 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
1712 (aarch64_legitimate_address_p): Likewise.
1713 (aarch64_print_address_internal): Take an aarch64_addr_query_type
1714 instead of an rtx code.
1715 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
1716 (aarch64_legitimate_address_hook_p): Likewise.
1717 (aarch64_print_ldpstp_address): Likewise.
1718 (aarch64_print_operand_address): Likewise.
1719 (aarch64_address_cost): Likewise.
1720 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
1721 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
1722 (aarch64_mem_pair_lanes_operand): Likewise.
1724 2017-12-20 Richard Biener <rguenther@suse.de>
1726 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
1727 update_stmt_if_modified.
1729 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
1731 PR tree-optimization/83491
1732 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
1733 before walking uses. Improve coding style and comments.
1735 2017-12-20 Tom de Vries <tom@codesourcery.com>
1737 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
1739 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1742 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
1744 (pass_ipa_free_fn_summary): Add small_p private data member,
1745 initialize to false in the ctor.
1746 (pass_ipa_free_fn_summary::clone,
1747 pass_ipa_free_fn_summary::set_pass_param,
1748 pass_ipa_free_fn_summary::gate): New methods.
1749 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
1750 | TODO_dump_symtab if small_p.
1751 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
1752 entry and add another instance of the pass with false parm after
1754 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
1755 ipa_free_fn_summary here.
1757 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
1759 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
1761 2017-12-20 Martin Sebor <msebor@redhat.com>
1764 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
1765 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
1766 (expand_builtin_strncmp): Same.
1768 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
1771 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
1772 allowed debug stmts before labels.
1773 (expand_gimple_basic_block): Likewise.
1774 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
1775 * gimple-iterator.h (gsi_after_labels): Likewise.
1776 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
1777 rename reused variable, and simplify using gsi_move_before.
1778 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
1779 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
1780 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
1781 (gimple_verify_flow_info, gimple_block_label): Likewise.
1782 (make_blocks): Move debug markers after adjacent labels.
1783 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
1784 allowed debug insns outside blocks.
1785 * df-scan.c (df_insn_delete): Likewise.
1786 * lra-constraints.c (update_ebb_live_info): Likewise.
1787 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
1788 (vt_initialize, delete_vta_debug_insns): Likewise.
1789 (reemit_marker_as_note): Drop BB parm. Adjust callers.
1791 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1792 Alan Hayward <alan.hayward@arm.com>
1793 David Sherwood <david.sherwood@arm.com>
1795 * poly-int-types.h (round_down_to_byte_boundary): New macro.
1796 (round_up_to_byte_boundary): Likewise.
1797 * expr.h (get_bit_range): Add temporary shim.
1798 * gimple-ssa-store-merging.c (store_operand_info): Change the
1799 bitsize, bitpos, bitregion_start and bitregion_end fields from
1800 unsigned HOST_WIDE_INT to poly_uint64.
1801 (merged_store_group): Likewise load_align_base.
1802 (compatible_load_p, compatible_load_p): Update accordingly.
1803 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1804 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1805 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1806 bitregion_start and bitregion_end as poly_uint64s rather than
1807 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1808 (handled_load): Take the bitsize, bitpos,
1809 bitregion_start and bitregion_end as poly_uint64s rather than
1810 unsigned HOST_WIDE_INTs.
1811 (pass_store_merging::process_store): Update call to
1812 mem_valid_for_store_merging.
1814 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1815 Alan Hayward <alan.hayward@arm.com>
1816 David Sherwood <david.sherwood@arm.com>
1818 * builtins.c (get_object_alignment_2): Track polynomial offsets
1819 and sizes. Update the alignment handling.
1821 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1822 Alan Hayward <alan.hayward@arm.com>
1823 David Sherwood <david.sherwood@arm.com>
1825 * tree.h (get_inner_reference): Add a version that returns the
1826 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1827 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1828 the case in which bitpos is not associated with the first byte.
1830 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1831 Alan Hayward <alan.hayward@arm.com>
1832 David Sherwood <david.sherwood@arm.com>
1834 * tree-affine.h (get_inner_reference_aff): Return the size as a
1836 * tree-affine.c (get_inner_reference_aff): Likewise.
1837 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1838 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
1840 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1841 Alan Hayward <alan.hayward@arm.com>
1842 David Sherwood <david.sherwood@arm.com>
1844 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1845 HOST_WIDE_INT rather than a poly_int64.
1847 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1848 Alan Hayward <alan.hayward@arm.com>
1849 David Sherwood <david.sherwood@arm.com>
1851 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1852 HOST_WIDE_INT to poly_int64_pod.
1853 (perform_symbolic_merge): Update accordingly.
1854 (bswap_replace): Likewise.
1856 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1857 Alan Hayward <alan.hayward@arm.com>
1858 David Sherwood <david.sherwood@arm.com>
1860 * tree-affine.h (aff_tree::offset): Change from widest_int
1862 (wide_int_ext_for_comb): Delete.
1863 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1864 constants as poly_widest_int rather than widest_int.
1865 (aff_combination_constant_multiple_p): Return the multiplier
1866 as a poly_widest_int.
1867 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1869 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1870 version static and add an overload for poly_widest_int.
1871 (aff_combination_const, aff_combination_add_cst)
1872 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1873 the constants as poly_widest_int rather than widest_int.
1874 (tree_to_aff_combination): Generalize INTEGER_CST case to
1876 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1877 (aff_combination_add_product, aff_combination_mult): Handle
1879 (aff_combination_constant_multiple_p): Return the multiplier
1880 as a poly_widest_int.
1881 * tree-predcom.c (determine_offset): Return the offset as a
1883 (split_data_refs_to_components, suitable_component_p): Update
1885 (valid_initializer_p): Update call to
1886 aff_combination_constant_multiple_p.
1887 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1888 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1889 as a poly_int64 rather than a HOST_WIDE_INT.
1890 (get_address_cost): Handle polynomial offsets.
1891 (iv_elimination_compare_lt): Likewise.
1892 (rewrite_use_nonlinear_expr): Likewise.
1894 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1895 Alan Hayward <alan.hayward@arm.com>
1896 David Sherwood <david.sherwood@arm.com>
1898 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1899 as a poly_int64_pod rather than a HOST_WIDE_INT.
1900 (get_addr_base_and_unit_offset): Likewise.
1901 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1902 (get_addr_base_and_unit_offset): Likewise.
1903 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1904 to poly_int64 in example.
1905 * fold-const.c (fold_binary_loc): Update call to
1906 get_addr_base_and_unit_offset.
1907 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1908 (maybe_canonicalize_mem_ref_addr): Likewise.
1909 (gimple_fold_stmt_to_constant_1): Likewise.
1910 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1912 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
1913 * match.pd: Likewise.
1914 * omp-low.c (lower_omp_target): Likewise.
1915 * tree-sra.c (build_ref_for_offset): Likewise.
1916 (build_debug_ref_for_model): Likewise.
1917 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
1918 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
1919 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
1920 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1921 (constant_pointer_difference): Likewise.
1922 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
1923 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
1924 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
1925 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1926 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
1927 (set_ssa_val_to): Likewise.
1928 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
1929 (maybe_diag_stxncpy_trunc): Likewise.
1930 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1931 * tree.c (build_simple_mem_ref_loc): Likewise.
1932 (array_at_struct_end_p): Likewise.
1934 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1935 Alan Hayward <alan.hayward@arm.com>
1936 David Sherwood <david.sherwood@arm.com>
1938 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
1939 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1940 (get_ref_base_and_extent_hwi): Declare.
1941 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
1942 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1943 (get_ref_base_and_extent_hwi): New function.
1944 * cfgexpand.c (expand_debug_expr): Update call to
1945 get_ref_base_and_extent.
1946 * dwarf2out.c (add_var_loc_to_decl): Likewise.
1947 * gimple-fold.c (get_base_constructor): Return the offset as a
1948 poly_int64_pod rather than a HOST_WIDE_INT.
1949 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
1950 * ipa-polymorphic-call.c
1951 (ipa_polymorphic_call_context::set_by_invariant)
1952 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
1953 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
1954 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
1955 rather than get_ref_base_and_extent.
1956 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
1957 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
1958 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
1960 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
1961 call to get_ref_base_and_extent.
1962 * tree-sra.c (create_access, get_access_for_expr): Likewise.
1963 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
1964 (stmt_kills_ref_p): Likewise.
1965 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1966 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
1968 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
1969 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
1970 when calling native_encode_expr.
1971 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
1972 call to get_ref_base_and_extent.
1973 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
1974 get_ref_base_and_extent.
1975 * var-tracking.c (track_expr_p): Likewise.
1977 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1978 Alan Hayward <alan.hayward@arm.com>
1979 David Sherwood <david.sherwood@arm.com>
1981 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
1982 HOST_WIDE_INT to poly_int64_pod.
1983 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
1984 polynomail parameter offsets.
1986 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1987 Alan Hayward <alan.hayward@arm.com>
1988 David Sherwood <david.sherwood@arm.com>
1990 * gengtype.c (main): Handle poly_int64_pod.
1991 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
1992 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
1993 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
1994 from HOST_WIDE_INT to poly_int64_pod.
1995 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
1996 (copy_cfa): New function.
1997 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
1998 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
1999 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
2000 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
2001 (initial_return_save): Treat offsets as poly_ints.
2002 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
2003 offset is nonconstant.
2004 (reg_save): Take the offset as a poly_int64. Fall back to
2005 DW_CFA_expression for nonconstant offsets.
2006 (queue_reg_save): Take the offset as a poly_int64.
2007 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
2009 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2010 Alan Hayward <alan.hayward@arm.com>
2011 David Sherwood <david.sherwood@arm.com>
2013 * rtl.h (operand_subword, operand_subword_force): Take the offset
2014 as a poly_uint64 an unsigned int.
2015 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
2017 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2018 Alan Hayward <alan.hayward@arm.com>
2019 David Sherwood <david.sherwood@arm.com>
2021 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
2022 'p' format code. Use INT_LIST rather than SUBREG as the example of
2023 a code with an XINT and an XEXP. Remove the implication that
2024 accessing an rtx field using XINT is expected to work.
2025 * rtl.def (SUBREG): Change format from "ei" to "ep".
2026 * rtl.h (rtunion::rt_subreg): New field.
2027 (XCSUBREG): New macro.
2028 (SUBREG_BYTE): Use it.
2029 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
2030 Update constructor accordingly.
2031 (subreg_shape::operator ==): Update accordingly.
2032 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
2033 than an unsigned int.
2034 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
2035 a poly_uint64 rather than an unsigned int.
2036 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
2037 than an unsigned int.
2038 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
2039 (subreg_size_highpart_offset): Return a poly_uint64 rather than
2040 an unsigned int. Take the sizes as poly_uint64s.
2041 (subreg_offset_from_lsb): Return a poly_uint64 rather than
2042 an unsigned int. Take the shift as a poly_uint64 rather than
2044 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
2045 as a poly_uint64 rather than an unsigned int.
2046 (simplify_subreg_regno): Likewise.
2047 (byte_lowpart_offset): Return the memory offset as a poly_int64
2049 (subreg_memory_offset): Likewise. Take the subreg offset as a
2050 poly_uint64 rather than an unsigned int.
2051 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
2052 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
2053 poly_uint64 rather than an unsigned int.
2054 * rtl.c (rtx_format): Describe 'p' in comment.
2055 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
2056 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
2057 offset as a poly_uint64 rather than an unsigned int.
2058 (byte_lowpart_offset): Return the memory offset as a poly_int64
2060 (subreg_memory_offset): Likewise. Take the subreg offset as a
2061 poly_uint64 rather than an unsigned int.
2062 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
2063 mode sizes as poly_uint64s rather than unsigned ints. Return a
2064 poly_uint64 rather than an unsigned int.
2065 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
2066 (copy_insn_1): Handle 'p'.
2067 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
2068 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
2069 an unsigned int. Return the shift in the same way.
2070 (subreg_lsb): Return the shift as a poly_uint64 rather than an
2072 (subreg_size_offset_from_lsb): Take the sizes and shift as
2073 poly_uint64s rather than unsigned ints. Return the offset as
2075 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
2076 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
2078 * rtlhash.c (add_rtx): Handle 'p'.
2079 * genemit.c (gen_exp): Likewise.
2080 * gengenrtl.c (type_from_format, gendef): Likewise.
2081 * gensupport.c (subst_pattern_match, get_alternatives_number)
2082 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
2083 (subst_dup): Likewise.
2084 * gengtype.c (adjust_field_rtx_def): Likewise.
2085 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
2086 (match_pattern_2): Likewise.
2087 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
2088 (rtx_test::subreg_field): New function.
2089 (operator ==, safe_to_hoist_p, transition_parameter_type)
2090 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
2091 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
2092 * genpeep.c (match_rtx): Likewise.
2093 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
2094 (rtx_writer::print_rtx_operand): Handle 'p'.
2095 (print_value): Handle SUBREG.
2096 * read-rtl.c (apply_int_iterator): Likewise.
2097 (rtx_reader::read_rtx_operand): Handle 'p'.
2098 * alias.c (rtx_equal_for_memref_p): Likewise.
2099 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
2100 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
2102 * calls.c (expand_call): Likewise.
2103 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
2104 (make_extraction, gen_lowpart_for_combine): Likewise.
2105 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
2107 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
2108 rather than an unsigned int. Treat subreg offsets as poly_ints.
2109 (exp_equiv_p): Handle 'p'.
2110 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
2111 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
2112 * dse.c (find_shift_sequence): Likewise.
2113 * dwarf2out.c (rtl_for_decl_location): Likewise.
2114 * expmed.c (extract_low_bits): Likewise.
2115 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
2116 (expand_expr_real_2): Likewise.
2117 * final.c (alter_subreg): Likewise.
2118 (leaf_renumber_regs_insn): Handle 'p'.
2119 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
2120 Treat subreg offsets as poly_ints.
2121 * fwprop.c (forward_propagate_and_simplify): Likewise.
2122 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
2123 * ira.c (get_subreg_tracking_sizes): Likewise.
2124 * ira-conflicts.c (go_through_subreg): Likewise.
2125 * ira-lives.c (process_single_reg_class_operands): Likewise.
2126 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
2127 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
2128 as a poly_uint64 rather than an unsigned int.
2129 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
2130 subreg offsets as poly_ints.
2131 * lra-constraints.c (operands_match_p): Handle 'p'.
2132 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
2133 * lra-spills.c (assign_mem_slot): Likewise.
2134 * postreload.c (move2add_valid_value_p): Likewise.
2135 * recog.c (general_operand, indirect_operand): Likewise.
2136 * regcprop.c (copy_value, maybe_mode_change): Likewise.
2137 (copyprop_hardreg_forward_1): Likewise.
2138 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
2139 (record_subregs_of_mode): Likewise.
2140 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
2141 * reload.c (operands_match_p): Handle 'p'.
2142 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
2143 * reload1.c (alter_reg, choose_reload_regs): Likewise.
2144 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
2145 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
2146 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
2147 (simplify_const_poly_int_tests<N>::run): Likewise.
2148 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
2149 a poly_uint64 rather than an unsigned int.
2150 * valtrack.c (debug_lowpart_subreg): Likewise.
2151 * var-tracking.c (var_lowpart): Likewise.
2152 (loc_cmp): Handle 'p'.
2154 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2155 Alan Hayward <alan.hayward@arm.com>
2156 David Sherwood <david.sherwood@arm.com>
2158 * ira.c (get_subreg_tracking_sizes): New function.
2159 (init_live_subregs): Take an integer size rather than a register.
2160 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
2161 to init_live_subregs.
2163 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2164 Alan Hayward <alan.hayward@arm.com>
2165 David Sherwood <david.sherwood@arm.com>
2167 * expr.c (store_constructor_field): Change bitsize from a
2168 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
2169 HOST_WIDE_INT to a poly_int64.
2170 (store_constructor): Change size from a HOST_WIDE_INT to
2172 (store_field): Likewise bitsize and bitpos.
2174 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2175 Alan Hayward <alan.hayward@arm.com>
2176 David Sherwood <david.sherwood@arm.com>
2178 * expmed.h (store_bit_field): Change bitregion_start and
2179 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
2180 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
2181 (store_bit_field_1, store_integral_bit_field, store_bit_field)
2182 (store_fixed_bit_field, store_split_bit_field): Likewise.
2183 * expr.c (store_constructor_field, store_field): Likewise.
2184 (optimize_bitfield_assignment_op): Likewise. Make the same change
2185 to bitsize and bitpos.
2186 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
2187 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
2188 same change in the constructor arguments.
2189 (get_best_mode): Change bitregion_start and bitregion_end from
2190 unsigned HOST_WIDE_INT to poly_uint64.
2191 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2192 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
2194 (bit_field_mode_iterator::next_mode): Update for new types
2195 of m_bitregion_start and m_bitregion_end.
2196 (get_best_mode): Change bitregion_start and bitregion_end from
2197 unsigned HOST_WIDE_INT to poly_uint64.
2199 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2200 Alan Hayward <alan.hayward@arm.com>
2201 David Sherwood <david.sherwood@arm.com>
2203 * rtl.h (simplify_gen_subreg): Add a temporary overload that
2204 accepts poly_uint64 offsets.
2205 * expmed.h (extract_bit_field): Take bitsize and bitnum as
2206 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2207 * expmed.c (lowpart_bit_field_p): Likewise.
2208 (extract_bit_field_as_subreg): New function, split out from...
2209 (extract_bit_field_1): ...here. Take bitsize and bitnum as
2210 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
2211 extractions, check that BITSIZE matches the size of the extracted
2212 value and that BITNUM is an exact multiple of that size.
2213 If all else fails, try forcing the value into memory if
2214 BITNUM is variable, and adjusting the address so that the
2215 offset is constant. Split the part that can only handle constant
2216 bitsize and bitnum out into...
2217 (extract_integral_bit_field): ...this new function.
2218 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
2219 rather than unsigned HOST_WIDE_INTs.
2221 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2222 Alan Hayward <alan.hayward@arm.com>
2223 David Sherwood <david.sherwood@arm.com>
2225 * expmed.h (store_bit_field): Take bitsize and bitnum as
2226 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2227 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
2228 that returns the byte size.
2229 (store_bit_field_1): Take bitsize and bitnum as
2230 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
2231 to simple_mem_bitfield_p. Split the part that can only handle
2232 constant bitsize and bitnum out into...
2233 (store_integral_bit_field): ...this new function.
2234 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
2235 than unsigned HOST_WIDE_INTs.
2236 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
2238 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2239 Alan Hayward <alan.hayward@arm.com>
2240 David Sherwood <david.sherwood@arm.com>
2242 * lra-int.h (lra_reg): Change offset from int to poly_int64.
2243 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
2245 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2246 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
2247 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
2248 offset as a poly_int64 rather than an int.
2249 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
2250 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
2251 * lra-constraints.c (equiv_address_substitution): Track offsets
2253 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
2254 (curr_insn_transform): Handle the new form of sp_offset.
2255 * lra-eliminations.c (lra_elim_table): Change previous_offset
2256 and offset from HOST_WIDE_INT to poly_int64.
2257 (print_elim_table, update_reg_eliminate): Update accordingly.
2258 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
2259 (get_elimination): Update accordingly.
2260 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
2261 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2262 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
2263 poly_int64 offsets generally.
2264 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
2265 (mark_not_eliminable, init_elimination): Update accordingly.
2266 (remove_reg_equal_offset_note): Return a bool and pass the new
2267 offset back by pointer as a poly_int64.
2268 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
2269 rather than a HOST_WIDE_INT.
2270 (do_remat): Track offsets poly_int64s.
2271 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
2273 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2274 Alan Hayward <alan.hayward@arm.com>
2275 David Sherwood <david.sherwood@arm.com>
2277 * rtl.h (mem_attrs): Add a default constructor. Change size and
2278 offset from HOST_WIDE_INT to poly_int64.
2279 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
2280 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
2281 (widen_memory_access): Take the sizes and offsets as poly_int64s
2282 rather than HOST_WIDE_INTs.
2283 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
2284 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
2286 (adjust_offset_for_component_ref): Change the offset from a
2287 HOST_WIDE_INT to a poly_int64.
2288 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
2289 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
2290 * dce.c (find_call_stack_args): Likewise.
2291 * dse.c (record_store): Likewise.
2292 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
2293 * print-rtl.c (rtx_writer::print_rtx): Likewise.
2294 * read-rtl-function.c (test_loading_mem): Likewise.
2295 * rtlanal.c (may_trap_p_1): Likewise.
2296 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
2297 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
2298 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
2299 (mem_attrs::mem_attrs): New function.
2300 (set_mem_attributes_minus_bitpos): Change bitpos from a
2301 HOST_WIDE_INT to poly_int64.
2302 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
2303 (clear_mem_offset, clear_mem_size, change_address)
2304 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
2305 initialize mem_attrs.
2306 (set_mem_offset, set_mem_size, adjust_address_1)
2307 (adjust_automodify_address_1, offset_address, widen_memory_access):
2308 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
2310 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2311 Alan Hayward <alan.hayward@arm.com>
2312 David Sherwood <david.sherwood@arm.com>
2314 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
2315 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
2316 rather than 0 to represent an unknown size. Assert that the size
2317 is known when the mode isn't BLKmode.
2318 (may_trap_p_1): Use -1 for unknown sizes.
2319 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
2321 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2322 Alan Hayward <alan.hayward@arm.com>
2323 David Sherwood <david.sherwood@arm.com>
2325 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
2326 to poly_int64. Update commentary for positions_needed.large.
2327 (read_info_type): Change offset and width from HOST_WIDE_INT
2329 (set_usage_bits): Likewise.
2330 (canon_address): Return the offset as a poly_int64 rather than
2331 a HOST_WIDE_INT. Use strip_offset_and_add.
2332 (set_all_positions_unneeded, any_positions_needed_p): Use
2333 positions_needed.large to track stores with non-constant widths.
2334 (all_positions_needed_p): Likewise. Take the offset and width
2335 as poly_int64s rather than ints. Assert that rhs is nonnull.
2336 (record_store): Cope with non-constant offsets and widths.
2337 Nullify the rhs of an earlier store if we can't tell which bytes
2339 (find_shift_sequence): Take the access_size and shift as poly_int64s
2341 (get_stored_val): Take the read_offset and read_width as poly_int64s
2342 rather than HOST_WIDE_INTs.
2343 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
2344 non-constant offsets and widths.
2346 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2347 Alan Hayward <alan.hayward@arm.com>
2348 David Sherwood <david.sherwood@arm.com>
2350 * inchash.h (inchash::hash::add_poly_int): New function.
2351 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
2352 Use poly_int64 rather than HOST_WIDE_INT.
2353 (ao_ref::max_size_known_p): New function.
2354 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
2355 rather than HOST_WIDE_INT.
2356 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
2357 to temporaries until its interface is adjusted to match.
2358 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
2359 (aliasing_component_refs_p, decl_refs_may_alias_p)
2360 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
2361 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
2362 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
2364 * alias.c (ao_ref_from_mem): Likewise.
2365 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
2366 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
2367 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
2368 (maybe_trim_complex_store, maybe_trim_constructor_store)
2369 (live_bytes_read, dse_classify_store): Likewise.
2370 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
2371 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
2372 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
2373 (vn_reference_lookup_3): Likewise.
2374 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
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 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
2381 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
2382 instead of ranges_overlap_p.
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 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
2389 sizes as poly_int64s rather than HOST_WIDE_INTs.
2391 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2392 Alan Hayward <alan.hayward@arm.com>
2393 David Sherwood <david.sherwood@arm.com>
2395 * gimple-fold.h (fold_ctor_reference): Take the offset and size
2396 as poly_uint64 rather than unsigned HOST_WIDE_INT.
2397 * gimple-fold.c (fold_ctor_reference): Likewise.
2399 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2400 Alan Hayward <alan.hayward@arm.com>
2401 David Sherwood <david.sherwood@arm.com>
2403 * target.def (dwarf_poly_indeterminate_value): New hook.
2404 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
2405 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
2406 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
2407 * doc/tm.texi: Regenerate.
2408 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
2409 offset as a poly_int64.
2410 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
2411 offset as a poly_int64.
2412 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
2413 Take the offset as a poly_int64.
2414 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
2415 (frame_pointer_fb_offset): Change to a poly_int64.
2416 (int_loc_descriptor): Take the offset as a poly_int64. Use
2417 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
2418 (based_loc_descr): Take the offset as a poly_int64.
2419 Use strip_offset_and_add to handle (plus X (const)).
2420 Use new_reg_loc_descr instead of an open-coded version of the
2421 previous implementation.
2422 (mem_loc_descriptor): Handle CONST_POLY_INT.
2423 (compute_frame_pointer_to_fb_displacement): Take the offset as a
2424 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
2426 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2427 Alan Hayward <alan.hayward@arm.com>
2428 David Sherwood <david.sherwood@arm.com>
2430 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
2431 (gen_rtx_REG_offset): Take the offset as a poly_int64.
2432 * inchash.h (inchash::hash::add_poly_hwi): New function.
2433 * gengtype.c (main): Register poly_int64.
2434 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
2435 offset as a poly_int.
2436 (reg_attr_hasher::equal): Use must_eq to compare offsets.
2437 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
2438 offset as a poly_int64.
2439 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
2440 * print-rtl.c (print_poly_int): New function.
2441 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
2443 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
2445 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
2446 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
2447 (vt_get_decl_and_offset): Return the offset as a poly_int64.
2448 Enforce track_offset_p for parts of a PARALLEL.
2449 (vt_add_function_parameter): Use const_offset for the final
2450 offset to track. Use get_tracked_reg_offset for the parts
2453 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2454 Alan Hayward <alan.hayward@arm.com>
2455 David Sherwood <david.sherwood@arm.com>
2457 * target.def (truly_noop_truncation): Take poly_uint64s instead of
2458 unsigned ints. Change default to hook_bool_puint64_puint64_true.
2459 * doc/tm.texi: Regenerate.
2460 * hooks.h (hook_bool_uint_uint_true): Delete.
2461 (hook_bool_puint64_puint64_true): Declare.
2462 * hooks.c (hook_bool_uint_uint_true): Delete.
2463 (hook_bool_puint64_puint64_true): New function.
2464 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
2465 instead of unsigned ints.
2466 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
2467 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
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 * optabs.h (expand_operand): Add an int_value field.
2474 (create_expand_operand): Add an int_value parameter and use it
2475 to initialize the new expand_operand field.
2476 (create_integer_operand): Replace with a declaration of a function
2477 that accepts poly_int64s. Move the implementation to...
2478 * optabs.c (create_integer_operand): ...here.
2479 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
2480 the mode preserves the value of int_value, instead of calling
2481 const_int_operand on the rtx. Use gen_int_mode to generate
2484 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2485 Alan Hayward <alan.hayward@arm.com>
2486 David Sherwood <david.sherwood@arm.com>
2488 * dumpfile.h (dump_dec): Declare.
2489 * dumpfile.c (dump_dec): New function.
2490 * pretty-print.h (pp_wide_integer): Turn into a function and
2491 declare a poly_int version.
2492 * pretty-print.c (pp_wide_integer): New function for poly_ints.
2494 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2495 Alan Hayward <alan.hayward@arm.com>
2496 David Sherwood <david.sherwood@arm.com>
2498 * doc/generic.texi (POLY_INT_CST): Document.
2499 * tree.def (POLY_INT_CST): New tree code.
2500 * treestruct.def (TS_POLY_INT_CST): New tree layout.
2501 * tree-core.h (tree_poly_int_cst): New struct.
2502 (tree_node): Add a poly_int_cst field.
2503 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
2504 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
2505 instead of a wide_int_ref.
2506 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
2508 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2509 instead of an unsigned HOST_WIDE_INT.
2510 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
2511 (ptrdiff_tree_p): Declare.
2512 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
2513 extern inline implementations if the target doesn't use POLY_INT_CST.
2514 (poly_int_tree_p): New function.
2515 (wi::unextended_tree): New class.
2516 (wi::int_traits <unextended_tree>): New override.
2517 (wi::extended_tree): Add a default constructor.
2518 (wi::extended_tree::get_tree): New function.
2519 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
2520 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
2521 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
2522 (wi::tree_to_poly_wide_ref): New typedefs.
2523 (wi::ints_for): Provide overloads for extended_tree and
2525 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
2526 (wi::to_wide): New functions.
2527 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
2528 * tree.c (poly_int_cst_hasher): New struct.
2529 (poly_int_cst_hash_table): New variable.
2530 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
2531 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
2533 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
2534 (init_ttree): Initialize poly_int_cst_hash_table.
2535 (build_int_cst, build_int_cst_type, build_invariant_address): Take
2536 a poly_int64 instead of a HOST_WIDE_INT.
2537 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2538 instead of an unsigned HOST_WIDE_INT.
2539 (wide_int_to_tree): Rename to...
2540 (wide_int_to_tree_1): ...this.
2541 (build_new_poly_int_cst, build_poly_int_cst): New functions.
2542 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
2543 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
2544 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
2545 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
2546 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
2548 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
2549 (streamer_read_tree_body): Likewise.
2550 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
2551 (streamer_write_tree_body): Likewise.
2552 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
2553 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
2554 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
2555 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
2556 * gimple-expr.h (is_gimple_constant): Likewise.
2557 * gimplify.c (maybe_with_size_expr): Likewise.
2558 * print-tree.c (print_node): Likewise.
2559 * tree-data-ref.c (data_ref_compare_tree): Likewise.
2560 * tree-pretty-print.c (dump_generic_node): Likewise.
2561 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2562 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
2563 * tree-vrp.c (compare_values_warnv): Likewise.
2564 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
2565 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
2566 (force_expr_to_var_cost): Likewise.
2567 * tree-ssa-loop.c (for_each_index): Likewise.
2568 * fold-const.h (build_invariant_address, size_int_kind): Take a
2569 poly_int64 instead of a HOST_WIDE_INT.
2570 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
2571 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
2573 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
2574 (int_const_binop_2): New function, split out from...
2575 (int_const_binop_1): ...here. Handle POLY_INT_CST.
2576 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
2577 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
2578 * gimple-ssa-strength-reduction.c (slsr_process_add)
2579 (slsr_process_mul): Check for INTEGER_CSTs before using them
2581 * stor-layout.c (bits_from_bytes): New function.
2582 (bit_from_pos): Use it.
2583 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
2584 by BITS_PER_UNIT to get the TYPE_SIZE.
2585 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
2586 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
2588 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2589 Alan Hayward <alan.hayward@arm.com>
2590 David Sherwood <david.sherwood@arm.com>
2592 * doc/rtl.texi (const_poly_int): Document. Also document the
2593 rtl sharing behavior.
2594 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
2595 * rtl.h (const_poly_int_def): New struct.
2596 (rtx_def::u): Add a cpi field.
2597 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
2598 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
2599 (wi::rtx_to_poly_wide_ref): New typedef
2600 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
2601 (poly_int_rtx_p): New functions.
2602 (trunc_int_for_mode): Declare a poly_int64 version.
2603 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2604 (immed_wide_int_const): Take a poly_wide_int_ref rather than
2606 (strip_offset): Declare.
2607 (strip_offset_and_add): New function.
2608 * rtl.def (CONST_POLY_INT): New rtx code.
2609 * rtl.c (rtx_size): Handle CONST_POLY_INT.
2610 (shared_const_p): Use poly_int_rtx_p.
2611 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
2613 (gen_int_shift_amount): Likewise.
2614 * emit-rtl.c (const_poly_int_hasher): New class.
2615 (const_poly_int_htab): New variable.
2616 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
2617 (const_poly_int_hasher::hash): New function.
2618 (const_poly_int_hasher::equal): Likewise.
2619 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
2620 (immed_wide_int_const): Rename to...
2621 (immed_wide_int_const_1): ...this and make static.
2622 (immed_wide_int_const): New function, taking a poly_wide_int_ref
2623 instead of a wide_int_ref.
2624 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
2625 (gen_lowpart_common): Handle CONST_POLY_INT.
2626 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
2627 * cselib.c (cselib_hash_rtx): Likewise.
2628 * dwarf2out.c (const_ok_for_output_1): Likewise.
2629 * expr.c (convert_modes): Likewise.
2630 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
2631 * rtlhash.c (add_rtx): Likewise.
2632 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
2633 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2634 Handle existing CONST_POLY_INT rtxes.
2635 * expmed.h (expand_shift): Take a poly_int64 instead of a
2637 * expmed.c (expand_shift): Likewise.
2638 * rtlanal.c (strip_offset): New function.
2639 (commutative_operand_precedence): Give CONST_POLY_INT the same
2640 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
2642 * rtl-tests.c (const_poly_int_tests): New struct.
2643 (rtl_tests_c_tests): Use it.
2644 * simplify-rtx.c (simplify_const_unary_operation): Handle
2646 (simplify_const_binary_operation): Likewise.
2647 (simplify_binary_operation_1): Fold additions of symbolic constants
2648 and CONST_POLY_INTs.
2649 (simplify_subreg): Handle extensions and truncations of
2651 (simplify_const_poly_int_tests): New struct.
2652 (simplify_rtx_c_tests): Use it.
2653 * wide-int.h (storage_ref): Add default constructor.
2654 (wide_int_ref_storage): Likewise.
2655 (trailing_wide_ints): Use GTY((user)).
2656 (trailing_wide_ints::operator[]): Add a const version.
2657 (trailing_wide_ints::get_precision): New function.
2658 (trailing_wide_ints::extra_size): Likewise.
2660 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2661 Alan Hayward <alan.hayward@arm.com>
2662 David Sherwood <david.sherwood@arm.com>
2664 * emit-rtl.h (gen_int_shift_amount): Declare.
2665 * emit-rtl.c (gen_int_shift_amount): New function.
2666 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
2668 * calls.c (shift_return_value): Likewise.
2669 * cse.c (fold_rtx): Likewise.
2670 * dse.c (find_shift_sequence): Likewise.
2671 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
2672 (expand_shift, expand_smod_pow2): Likewise.
2673 * lower-subreg.c (shift_cost): Likewise.
2674 * optabs.c (expand_superword_shift, expand_doubleword_mult)
2675 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
2676 (expand_vec_perm_var): Likewise.
2677 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2678 (simplify_binary_operation_1): Likewise.
2679 * combine.c (try_combine, find_split_point, force_int_to_mode)
2680 (simplify_shift_const_1, simplify_shift_const): Likewise.
2681 (change_zero_ext): Likewise. Use simplify_gen_binary.
2683 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2685 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
2687 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2689 * doc/invoke.texi (ARM Options): Document accepted extension options
2690 for -march=armv8.3-a.
2692 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
2695 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
2696 or --with-float=softfp, set the default CPU to arm10e.
2698 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
2700 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
2701 * config/visium/predicates.md (const_shift_operand): Likewise.
2702 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
2703 (visium_legitimize_reload_address): Likewise.
2705 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2708 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
2710 2017-12-20 Jakub Jelinek <jakub@redhat.com>
2713 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
2715 2017-12-20 Julia Koval <julia.koval@intel.com>
2717 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
2718 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
2719 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
2720 * config.gcc: Include vpclmulqdqintrin.h.
2721 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
2722 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
2724 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
2725 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
2726 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
2727 * config/i386/i386.c (isa2_opts): Add -mcx16.
2728 (isa_opts): Add -mpclmulqdq, remove -mcx16.
2729 (ix86_option_override_internal): Move mcx16 to flags2.
2730 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
2731 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
2732 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
2733 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
2734 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
2735 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
2736 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
2737 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
2738 * doc/invoke.texi: Add -mvpclmulqdq.
2740 2017-12-20 Tom de Vries <tom@codesourcery.com>
2743 * config/i386/i386.c (ix86_static_chain): Move
2744 DECL_STATIC_CHAIN test ...
2745 * calls.c (rtx_for_static_chain): ... here. New function.
2746 * calls.h (rtx_for_static_chain): Declare.
2747 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
2748 instead of targetm.calls.static_chain.
2749 * df-scan.c (df_get_entry_block_def_set): Same.
2751 2017-12-19 Tom de Vries <tom@codesourcery.com>
2753 PR tree-optimization/83493
2754 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
2757 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2759 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
2760 inform with hardcoded english plural handling.
2762 2017-12-18 Jeff Law <law@redhat.com>
2764 PR tree-optimization/83477
2765 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
2766 a non-virtual PHI, always push a new range.
2768 2017-12-19 Martin Sebor <msebor@redhat.com>
2771 * builtins.c (compute_objsize): Handle non-constant offsets.
2773 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2775 PR tree-optimization/83444
2776 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
2777 character load case, if get_stridx on MEM_REF's operand doesn't
2778 look usable, retry with get_addr_stridx.
2780 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
2783 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
2784 (variable_tracking_main_1): Keep markers even when VTA fails.
2787 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
2788 even if there are markers before it.
2789 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
2791 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2793 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
2794 typical order conditions.
2795 * sel-sched.c (extract_new_fences_from): Likewise.
2796 * config/visium/constraints.md (J, K, L): Likewise.
2797 * config/visium/predicates.md (const_shift_operand): Likewise.
2798 * config/visium/visium.c (visium_legitimize_address,
2799 visium_legitimize_reload_address): Likewise.
2800 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2801 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2802 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2803 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2804 SET_DUMP_DETAIL): Likewise.
2805 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2806 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2807 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2808 avr_set_core_architecture, avr_set_current_function,
2809 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2810 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2811 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2812 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2813 avr_map_decompose, avr_fold_builtin): Likewise.
2814 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2815 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2816 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2817 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2818 * config/m32c/m32c.c (m32c_conditional_register_usage,
2819 m32c_address_cost): Likewise.
2820 * config/m32c/predicates.md (shiftcount_operand,
2821 longshiftcount_operand): Likewise.
2822 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2823 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2824 can_use_cdx_ldstw): Likewise.
2825 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2826 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2828 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2829 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2830 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2831 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2832 * config/vax/vax.c (adjacent_operands_p): Likewise.
2833 * config/ft32/constraints.md (L, b, KA): Likewise.
2834 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2836 * cfgexpand.c (expand_stack_alignment): Likewise.
2837 * gcse.c (insert_expr_in_table): Likewise.
2838 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2839 * cgraphunit.c (cgraph_node::expand): Likewise.
2840 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2841 * emit-rtl.c (add_insn): Likewise.
2842 * input.c (dump_location_info): Likewise.
2843 * passes.c (NEXT_PASS): Likewise.
2844 * read-rtl-function.c (parse_note_insn_name,
2845 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2847 * sched-rgn.c (sched_rgn_init): Likewise.
2848 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2849 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2850 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2852 * explow.c (eliminate_constant_term): Likewise.
2853 * final.c (leaf_renumber_regs_insn): Likewise.
2854 * cfgrtl.c (print_rtl_with_bb): Likewise.
2855 * genhooks.c (emit_init_macros): Likewise.
2856 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2857 * tree-data-ref.c (conflict_fn): Likewise.
2858 * selftest.c (assert_streq): Likewise.
2859 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2860 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2861 fold_binary_loc, multiple_of_p): Likewise.
2862 * reload.c (push_reload, find_equiv_reg): Likewise.
2863 * et-forest.c (et_nca, et_below): Likewise.
2864 * dbxout.c (dbxout_symbol_location): Likewise.
2865 * reorg.c (relax_delay_slots): Likewise.
2866 * dojump.c (do_compare_rtx_and_jump): Likewise.
2867 * gengtype-parse.c (type): Likewise.
2868 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2869 simplify_const_relational_operation): Likewise.
2870 * reload1.c (do_output_reload): Likewise.
2871 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2872 * gengtype.c (type_for_name): Likewise.
2873 * gimple-ssa-sprintf.c (format_directive): Likewise.
2875 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2878 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2879 accessing it. Adjust comment.
2881 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2884 * predict.c (zero_one_minusone): New function.
2885 (apply_return_prediction): Avoid return prediction for functions
2886 returning only -1, 0 and 1 values, unless they only return -1 and 0
2889 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2891 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2892 fall-through warning.
2893 (arc_compute_frame_size): Remove unused variables.
2894 (arc_print_operand): Fix fprintif format.
2895 (arc_can_follow_jump): Clean fall-through warning.
2897 2017-12-19 Marek Polacek <polacek@redhat.com>
2900 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2903 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2905 * config/arc/arc.c (overriderregs): New variable.
2906 (arc_override_options): Track fixed/call saved/call options.
2907 (arc_conditional_register_usage): Check against overrideregs
2908 variable whenever we change register properties.
2910 2017-12-19 Nathan Sidwell <nathan@acm.org>
2912 * opts.c (finish_options): Don't prefix dump_base_name if it
2913 already contains directories.
2915 2017-12-19 Martin Liska <mliska@suse.cz>
2917 PR rtl-optimization/82675
2918 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
2919 more element in sbitmap.
2921 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
2923 * gimplify.c (gimplify_expr): Use error_operand_p.
2925 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
2928 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
2931 2017-12-19 Daniel Cederman <cederman@gaisler.com>
2933 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
2934 the jump is to a label.
2936 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2938 PR tree-optimization/83444
2939 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
2944 * ipa-inline.c (flatten_remove_node_hook): New function.
2945 (ipa_inline): Keep only nodes with flatten attribute at the end of
2946 the array in the order from ipa_reverse_postorder, only walk that
2947 portion of array for flattening, if there is more than one such
2948 node, temporarily register a removal hook and ignore removed nodes.
2950 PR tree-optimization/80631
2951 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
2952 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
2953 IFN_REDUC_MAX or IFN_REDUC_MIN.
2955 2017-12-18 Jeff Law <law@redhat.com>
2957 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
2958 record symbolic equivalences from backedges in the CFG.
2961 2017-11-19 Jeff Law <law@redhat.com>
2963 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2964 of degenerates resulting from ignoring an edge.
2966 2017-12-18 Martin Sebor <msebor@redhat.com>
2969 PR tree-optimization/78450
2970 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
2971 (handle_builtin_strlen): Call it.
2973 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
2975 PR rtl-optimization/83424
2976 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
2978 2017-12-18 Marek Polacek <polacek@redhat.com>
2981 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
2982 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
2985 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2987 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
2988 (umaddsidi4, umaddsidi_split): Likewise.
2990 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2992 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
2995 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2998 * config/s390/s390.c (s390_option_override): Avoid strncpy.
3000 2017-12-18 Richard Biener <rguenther@suse.de>
3002 PR tree-optimization/81877
3003 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
3004 (outermost_indep_loop): Adjust.
3005 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
3006 (can_sm_ref_p): Adjust.
3008 2017-12-18 Richard Biener <rguenther@suse.de>
3011 * tree.c (array_at_struct_end_p): Return true if the underlying
3012 object has space for at least one element in excess of what
3013 the array domain specifies.
3015 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3017 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
3018 example to avoid overfull hbox.
3019 * doc/invoke.texi (Option Summary): Add missing @gol.
3020 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
3023 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3024 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3026 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
3027 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
3028 (AArch64 Options): Document -mverbose-cost-dump.
3029 (ARM Options): Likewise, plus -mflip-thumb.
3031 2017-12-17 Martin Sebor <msebor@redhat.com>
3034 * gimple-ssa-warn-restrict.c
3035 (builtin_memref::offset_out_of_bounds): Correct the handling of
3038 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3040 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
3042 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
3044 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
3046 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
3048 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
3051 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
3053 * doc/invoke.texi: Fix some typos.
3055 2017-12-16 Martin Sebor <msebor@redhat.com>
3057 PR tree-optimization/78918
3058 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
3059 * builtins.c (check_sizes): Rename...
3060 (check_access): ...to this. Rename function arguments for clarity.
3061 (check_memop_sizes): Adjust names.
3062 (expand_builtin_memchr, expand_builtin_memcpy): Same.
3063 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
3064 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
3065 (check_strncat_sizes, expand_builtin_strncat): Same.
3066 (expand_builtin_strncpy, expand_builtin_memset): Same.
3067 (expand_builtin_bzero, expand_builtin_memcmp): Same.
3068 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
3069 (maybe_emit_sprintf_chk_warning): Same.
3070 (expand_builtin_strcpy): Adjust.
3071 (expand_builtin_stpcpy): Same.
3072 (expand_builtin_with_bounds): Detect out-of-bounds accesses
3073 in pointer-checking forms of memcpy, memmove, and mempcpy.
3074 (gcall_to_tree_minimal, max_object_size): Define new functions.
3075 * builtins.h (max_object_size): Declare.
3076 * calls.c (alloc_max_size): Call max_object_size instead of
3077 hardcoding ssizetype limit.
3078 (get_size_range): Handle new argument.
3079 * calls.h (get_size_range): Add a new argument.
3080 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
3081 * doc/invoke.texi (-Wrestrict): Adjust, add example.
3082 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
3084 (gimple_fold_builtin_memory_chk): Same.
3085 (gimple_fold_builtin_stxcpy_chk): New function.
3086 * gimple-ssa-warn-restrict.c: New source.
3087 * gimple-ssa-warn-restrict.h: New header.
3088 * gimple.c (gimple_build_call_from_tree): Propagate location.
3089 * passes.def (pass_warn_restrict): Add new pass.
3090 * tree-pass.h (make_pass_warn_restrict): Declare.
3091 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
3093 (handle_builtin_strcat): Same.
3094 (strlen_optimize_stmt): Rename...
3095 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
3096 stpncpy, strncpy, and their checking forms.
3098 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
3100 PR rtl-optimization/82849
3101 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
3102 and get_max_loop_iterations_int.
3104 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3105 Alan Hayward <alan.hayward@arm.com>
3106 David Sherwood <david.sherwood@arm.com>
3108 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
3109 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
3110 size as a poly_uint64.
3111 (mode_for_vector, mode_for_int_vector): Take the number of vector
3112 elements as a poly_uint64.
3113 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
3114 size as a poly_uint64.
3115 (mode_for_vector, mode_for_int_vector): Take the number of vector
3116 elements as a poly_uint64.
3118 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3119 Alan Hayward <alan.hayward@arm.com>
3120 David Sherwood <david.sherwood@arm.com>
3122 * machmode.h (MACRO_MODE): New macro.
3123 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
3124 * targhooks.c (default_libcall_value, default_secondary_reload)
3125 (default_memory_move_cost, default_register_move_cost)
3126 (default_class_max_nregs): Likewise.
3128 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3129 Alan Hayward <alan.hayward@arm.com>
3130 David Sherwood <david.sherwood@arm.com>
3132 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
3133 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
3135 * genautomata.c (main): Likewise.
3136 * genconditions.c (write_header): Likewise.
3137 * genemit.c (main): Likewise.
3138 * genextract.c (print_header): Likewise.
3139 * genopinit.c (main): Likewise.
3140 * genoutput.c (output_prologue): Likewise.
3141 * genpeep.c (main): Likewise.
3142 * genpreds.c (write_insn_preds_c): Likewise.
3143 * genrecog.c (writer_header): Likewise.
3144 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
3145 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
3146 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
3147 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
3148 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
3149 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
3150 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
3151 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
3152 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
3153 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
3154 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
3155 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
3156 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
3157 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
3158 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
3159 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
3160 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
3161 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
3162 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
3163 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
3164 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
3165 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
3166 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
3167 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
3168 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
3169 * config/darwin.c (IN_TARGET_CODE): Likewise.
3170 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
3171 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
3172 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
3173 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
3174 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
3175 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
3176 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
3177 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
3178 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
3179 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
3180 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
3181 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
3182 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
3183 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
3184 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
3185 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
3186 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
3187 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
3188 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
3189 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
3190 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
3191 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
3192 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
3193 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
3194 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
3195 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
3196 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
3197 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
3198 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
3199 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
3200 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
3201 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
3202 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
3203 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
3204 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
3205 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
3206 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
3207 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
3208 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
3209 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
3210 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
3211 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
3212 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
3213 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
3214 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
3215 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
3216 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
3217 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
3218 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
3219 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
3220 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
3221 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
3222 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
3223 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
3224 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
3225 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
3226 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
3227 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
3228 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
3229 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
3230 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3231 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
3232 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
3233 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
3234 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
3235 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
3236 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
3237 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
3238 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
3239 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
3240 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
3241 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3242 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
3243 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
3244 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
3245 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
3246 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
3247 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
3248 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
3249 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
3250 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
3251 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
3252 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
3253 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
3254 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
3255 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
3256 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
3257 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
3258 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
3259 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
3260 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
3261 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
3262 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
3263 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
3264 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
3265 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
3266 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
3267 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
3268 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
3269 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
3270 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
3271 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
3272 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
3273 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
3274 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
3275 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
3277 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3278 Alan Hayward <alan.hayward@arm.com>
3279 David Sherwood <david.sherwood@arm.com>
3281 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
3284 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3285 Alan Hayward <alan.hayward@arm.com>
3286 David Sherwood <david.sherwood@arm.com>
3288 * doc/generic.texi (VEC_SERIES_EXPR): Document.
3289 * doc/md.texi (vec_series@var{m}): Document.
3290 * tree.def (VEC_SERIES_EXPR): New tree code.
3291 * tree.h (build_vec_series): Declare.
3292 * tree.c (build_vec_series): New function.
3293 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
3294 * tree-pretty-print.c (dump_generic_node): Likewise.
3295 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
3296 * tree-inline.c (estimate_operator_cost): Likewise.
3297 * expr.c (expand_expr_real_2): Likewise.
3298 * optabs-tree.c (optab_for_tree_code): Likewise.
3299 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
3300 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
3301 * expmed.c (make_tree): Handle VEC_SERIES.
3302 * optabs.def (vec_series_optab): New optab.
3303 * optabs.h (expand_vec_series_expr): Declare.
3304 * optabs.c (expand_vec_series_expr): New function.
3305 * tree-vect-generic.c (expand_vector_operations_1): Check that
3306 the operands also have vector type.
3308 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3309 Alan Hayward <alan.hawyard@arm.com>
3310 David Sherwood <david.sherwood@arm.com>
3312 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
3313 (VEC_COND_EXPR): Add missing @tindex.
3314 * doc/md.texi (vec_duplicate@var{m}): Document.
3315 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
3316 * tree.c (build_vector_from_val): Add stubbed-out handling of
3317 variable-length vectors, using VEC_DUPLICATE_EXPR.
3318 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
3319 * cfgexpand.c (expand_debug_expr): Likewise.
3320 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3321 * tree-inline.c (estimate_operator_cost): Likewise.
3322 * tree-pretty-print.c (dump_generic_node): Likewise.
3323 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
3324 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
3325 (test_vec_duplicate_folding): New function.
3326 (fold_const_c_tests): Call it.
3327 * optabs.def (vec_duplicate_optab): New optab.
3328 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
3329 * optabs.h (expand_vector_broadcast): Declare.
3330 * optabs.c (expand_vector_broadcast): Make non-static. Try using
3331 vec_duplicate_optab.
3332 * expr.c (store_constructor): Try using vec_duplicate_optab for
3334 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
3336 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
3339 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
3340 div/mod latencies a bit.
3342 2017-12-15 Jeff Law <law@redhat.com>
3344 PR tree-optimization/36550
3345 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
3346 (mark_threaded_blocks): Rewrite code to avoid block copying when
3347 optimizing for size. Don't pessimize blocks which will be
3348 copied, but all the statements will be dead.
3350 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
3352 PR tree-optimization/81165
3353 * tree-ssa-threadupdate.c (uses_in_bb): New.
3354 (estimate_threading_killed_stmts): New.
3355 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
3356 * tree-ssa-threadedge.c
3357 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
3360 2017-12-15 Jeff Law <law@redhat.com>
3362 PR tree-optimization/83410
3363 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
3364 threads when parallelizing loops.
3366 2017-12-15 Jakub Jelinek <jakub@redhat.com>
3368 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
3370 * config/alpha/alpha.c (vms_attribute_table): Swap
3371 affects_type_identity and handler fields, adjust comments.
3372 * config/mips/mips.c (mips_attribute_table): Likewise.
3373 * config/visium/visium.c (visium_attribute_table): Likewise.
3374 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
3375 * config/microblaze/microblaze.c (microblaze_attribute_table):
3377 * config/spu/spu.c (spu_attribute_table): Likewise.
3378 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
3379 * config/arc/arc.c (arc_attribute_table): Likewise.
3380 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
3381 * config/v850/v850.c (v850_handle_interrupt_attribute,
3382 v850_handle_data_area_attribute): Formatting fixes.
3383 (v850_attribute_table): Swap affects_type_identity and handler
3384 fields, adjust comments.
3385 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
3386 * config/arm/arm.c (arm_attribute_table): Likewise.
3387 * config/avr/avr.c (avr_attribute_table): Likewise.
3388 * config/s390/s390.c (s390_attribute_table): Likewise.
3389 * config/sh/sh.c (sh_attribute_table): Likewise.
3390 * config/i386/i386.c (ix86_handle_cconv_attribute,
3391 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
3392 (ix86_attribute_table): Swap affects_type_identity and handler
3393 fields, adjust comments.
3394 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3395 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
3396 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
3397 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
3398 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
3399 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3400 * config/rx/rx.c (rx_attribute_table): Likewise.
3401 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
3402 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
3403 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
3404 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
3405 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3406 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
3407 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
3408 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
3409 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3410 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3411 * doc/plugins.texi (user_attr): Likewise. Add NULL for
3413 * attribs.c (empty_attribute_table): Swap affects_type_identity and
3415 (register_scoped_attributes, decl_attributes): Formatting fixes.
3417 PR tree-optimization/83269
3418 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
3419 subtraction in arg0's type if type is signed and arg0 is unsigned.
3423 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
3424 convert? on @0 instead of convert. Check type of @1, not @0.
3425 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
3426 convert? on @0 instead of convert. Check type of @1, not @0.
3427 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
3428 only optimize if either both @1 and @2 types are narrower
3429 precision, or both are wider or equal precision, and in the former
3430 case only if both have undefined overflow.
3432 2017-12-15 Richard Biener <rguenther@suse.de>
3435 * internal-fn.def (IFN_NOP): Add.
3436 * internal-fn.c (expand_NOP): Do nothing.
3437 * lto-streamer-in.c (input_function): Instead of removing
3438 sanitizer calls replace them with IFN_NOP calls.
3440 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
3441 Alan Hayward <alan.hayward@arm.com>
3442 David Sherwood <david.sherwood@arm.com>
3444 * dse.c (store_info, read_info_type): Replace begin and end with
3446 (print_range): New function.
3447 (set_all_positions_unneeded, any_positions_needed_p)
3448 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
3450 (record_store): Likewise. Optimize the case in which all positions
3452 (get_stored_val): Replace read_begin and read_end with read_offset
3454 (replace_read): Update call accordingly.
3456 2017-12-15 Bin Cheng <bin.cheng@arm.com>
3458 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
3459 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
3460 (loop_cand::loop_cand): Initialize above members.
3461 (loop_cand::supported_operations): Delete.
3462 (loop_cand::can_interchange_p): Inline above function.
3463 (loop_cand::classify_simple_reduction): Record number of constant
3464 initialized simple reductions.
3465 (should_interchange_loops): New parameters. Check stmt cost of loops
3467 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
3468 Update call to should_interchange_loops.
3469 (should_interchange_loop_nest): Update call to
3470 should_interchange_loops.
3472 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
3475 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
3476 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
3477 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
3479 2017-12-15 Julia Koval <julia.koval@intel.com>
3481 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
3482 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
3484 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
3485 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
3486 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
3488 2017-12-15 Julia Koval <julia.koval@intel.com>
3490 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
3491 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
3492 * config/i386/sse.md (vaesenc_<mode>): New pattern.
3493 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
3494 _mm_aesenc_epi128): New intrinsics.
3496 2017-12-15 Julia Koval <julia.koval@intel.com>
3498 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
3499 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
3501 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
3502 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
3503 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
3505 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3507 * gimple-ssa-strength-reduction.c (analyze_increments):
3508 Distinguish replacement costs for constant strides from those for
3511 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3513 * var-tracking.c (variable_tracking_main_1): Formatting fix.
3515 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
3517 * doc/invoke.texi: Document -Wcast-function-type.
3518 * recog.h (stored_funcptr): Change signature.
3519 * tree-dump.c (dump_node): Avoid warning.
3520 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
3522 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
3525 * gimple-fold.c (get_range_strlen): Add the handling of non-member
3528 2017-12-14 David Malcolm <dmalcolm@redhat.com>
3530 PR tree-optimization/83312
3531 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
3532 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
3533 "val" param, and to cope with arbitrary basic blocks.
3534 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
3535 handle NULL_TREE for "val", dropping "bb" param.
3536 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
3537 drop "bb" param. Handle NULL_TREE for "val".
3538 (find_case_label_for_value): Make "switch_stmt" param const.
3539 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
3541 (vrp_prop::check_all_array_refs): Reimplement as...
3542 (check_array_bounds_dom_walker::before_dom_children): ...this new
3543 vfunc. Replace linear search through BB block list, excluding
3544 those with non-executable in-edges via dominator walk.
3546 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3548 * config/arm/arm.opt (mverbose-cost-dump): New option.
3549 * config/arm/arm.c (arm_rtx_costs): Use it.
3551 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
3554 * reload1.c (emit_input_reload_insns): Skip debug markers.
3556 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
3558 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
3562 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
3563 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
3564 * config/sh/sh.c: Skip debug insns besides notes.
3565 * config/sh/sh.md: Likewise.
3566 * config/sh/sh_treg_combine.cc: Likewise.
3567 * config/sh/sync.md: Likewise.
3569 2017-12-14 Tom de Vries <tom@codesourcery.com>
3571 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
3572 Add item for weak_undefined.
3574 2017-12-14 Richard Biener <rguenther@suse.de>
3576 PR tree-optimization/67842
3577 * sese.h (bb_in_region): Remove #if 0'ed code.
3579 2017-12-14 Richard Biener <rguenther@suse.de>
3581 PR tree-optimization/83326
3582 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3583 may_be_zero parameter and handle it by not marking the first
3584 peeled copy as not exiting the loop.
3585 (try_peel_loop): Likewise.
3586 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
3587 to handle the case of constant or zero iterations and perform
3588 loop header copying on-the-fly.
3590 2017-12-14 Richard Biener <rguenther@suse.de>
3592 PR tree-optimization/83418
3593 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
3594 Instead of asserting we don't get unfolded comparisons deal with them.
3596 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3599 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
3600 basic blocks. Assert debug bind insns don't appear outside of bbs,
3601 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
3604 PR tree-optimization/83198
3605 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
3606 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
3607 value if arg is a REAL_CST with incompatible type.
3609 2017-12-14 Sudakshina Das <sudi.das@arm.com>
3610 Bin Cheng <bin.cheng@arm.com>
3613 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
3615 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
3618 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3620 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
3621 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
3623 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
3624 Alan Hayward <alan.hayward@arm.com>
3625 David Sherwood <david.sherwood@arm.com>
3627 * poly-int.h: New file.
3628 * poly-int-types.h: Likewise.
3629 * coretypes.h: Include them.
3630 (POLY_INT_CONVERSION): Define.
3631 * target.def (estimated_poly_value): New hook.
3632 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
3633 * doc/tm.texi: Regenerate.
3634 * doc/poly-int.texi: New file.
3635 * doc/gccint.texi: Include it.
3636 * doc/rtl.texi: Describe restrictions on subreg modes.
3637 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
3638 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
3639 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
3640 * targhooks.h (default_estimated_poly_value): Declare.
3641 * targhooks.c (default_estimated_poly_value): New function.
3642 * target.h (estimated_poly_value): Likewise.
3643 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
3644 (wi::unary_traits): Delete.
3645 (wi::binary_traits::signed_shift_result_type): Define for
3646 offset_int << HOST_WIDE_INT, etc.
3647 (generic_wide_int::operator <<=): Define for all types and use
3648 wi::lshift instead of <<.
3649 (wi::hwi_with_prec): Add a default constructor.
3650 (wi::ints_for): New class.
3651 (operator <<): Define for all wide-int types.
3652 (operator /): New function.
3653 (operator %): Likewise.
3654 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
3655 (ASSERT_MAYBE_NE_AT): New macros.
3657 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
3658 Dominik Vogt <vogt@linux.vnet.ibm.com>
3661 * emit-rtl.c (init_emit): Remove ??? comment.
3662 * explow.c (get_dynamic_stack_size): Take known alignment of stack
3663 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
3664 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
3665 alignment of 3 virtual registers to BITS_PER_WORD.
3667 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
3669 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
3671 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
3672 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
3674 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3677 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
3679 2017-12-13 Jakub Jelinek <jakub@redhat.com>
3681 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
3682 with the exception of debug begin stmt markers appear before
3686 * final.c (rest_of_handle_final): Call variable_tracking_main only
3687 if !flag_var_tracking.
3689 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3690 Jakub Jelinek <jakub@redhat.com>
3694 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
3695 labels debug stmts that can only appear after labels.
3697 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
3699 PR rtl-optimization/82398
3700 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
3701 EXPR_USEFULNESS in priority comparison.
3703 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
3705 PR rtl-optimization/83393
3706 * combine.c (move_deaths): If reg_stat points to a too new insn in
3707 last_death, do not use it: find the proper insn instead.
3709 2017-12-12 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3712 * config/rs6000/rs6000-string.c (expand_block_compare,
3713 expand_strn_compare): Fix set_mem_size() calls.
3715 2017-12-12 Jeff Law <law@redhat.com>
3717 PR tree-optimization/83298
3718 PR tree-optimization/83362
3719 PR tree-optimization/83383
3720 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
3721 push_value_range a public interface. Add new argument to
3722 record_ranges_from_stmt.
3723 * gimple-ssa-evrp-analyze.c
3724 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
3725 Update comments. Handle recording temporary equivalences.
3726 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
3727 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
3728 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
3729 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
3730 gimple-ssa-evrp-analyze.h.
3731 (record_temporary_equivalences_from_phis): Add new argument. When
3732 the PHI arg is an SSA_NAME, set the result's range to the range
3734 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
3735 from statements too.
3736 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
3737 Pass it down to children as needed.
3738 (thread_outgoing_edges): Likewise.
3739 (thread_across_edge): Likewise. Push/pop range state as needed.
3740 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
3742 2017-12-12 Julia Koval <julia.koval@intel.com>
3744 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
3745 (PTA_CANNONLAKE): Remove PTA_CLWB.
3747 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3750 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
3751 (vcondv2div2df): Likewise.
3752 (vconduv2dfv2di): Likewise.
3754 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3756 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
3757 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
3758 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
3759 ATTR_MATHFN_FPROUNDING_ERRNO.
3761 2017-12-12 Richard Biener <rguenther@suse.de>
3763 PR tree-optimization/83385
3764 * graphite-scop-detection.c (get_order, order): Remove.
3765 (bb_to_rpo): New global.
3767 (build_scops): Sort pbbs in RPO order.
3769 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
3771 * combine.c (simplify_set): Do not transform subregs to zero_extends
3772 if the destination mode is a vector mode.
3774 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3776 PR tree-optimization/83359
3777 * tree-cfg.h (fold_loop_internal_call): Declare.
3778 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
3779 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
3780 (find_loop_dist_alias): New function.
3781 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
3782 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
3785 PR tree-optimization/80631
3786 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
3787 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
3788 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
3789 hardcoding zero as the value if COND_EXPR is never true. For
3790 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
3791 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
3792 hardcoding MAX_EXPR as the reduction operation.
3793 (is_nonwrapping_integer_induction): Allow negative step.
3794 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
3795 vect_create_epilog_for_reduction, if no value is suitable, don't
3796 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
3798 2017-12-12 Richard Biener <rguenther@suse.de>
3800 PR tree-optimization/81889
3801 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
3802 range info from the non-wrapping IV instead of just the range
3805 2017-12-12 Julia Koval <julia.koval@intel.com>
3807 * config.gcc: Add vaesintrin.h.
3808 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3810 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3811 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3813 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3814 * config/i386/immintrin.h: Include vaesintrin.h.
3815 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3816 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3817 _mm_aesdec_epi128): New intrinsics.
3819 2017-12-12 Julia Koval <julia.koval@intel.com>
3821 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3822 OPTION_MASK_ISA_VAES_UNSET): New.
3823 (ix86_handle_option): Handle -mvaes.
3824 * config/i386/cpuid.h: Define bit_VAES.
3825 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3826 * config/i386/i386-c.c (__VAES__): New.
3827 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3828 (ix86_valid_target_attribute_inner_p): Ditto.
3829 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3830 * config/i386/i386.opt: Add -mvaes.
3831 * doc/invoke.texi: Ditto.
3833 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3835 * debug.h (gcc_debug_hooks): Add inline_entry.
3836 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3837 * debug.c (do_nothing_debug_hooks): Likewise.
3838 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3839 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3840 (dwarf2_lineno_debug_hooks): Likewise.
3842 * common.opt (gstatement-frontiers): New, setting
3843 debug_nonbind_markers_p.
3844 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
3845 * toplev.c (process_options): Autodetect value for debug statement
3847 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3848 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3850 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3851 markers. Integrate source bind into debug stmt expand loop.
3852 (pass_expand::execute): Check debug marker limit. Avoid deep
3853 TER and expand debug locations for debug bind insns only.
3854 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3856 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3857 * final.c (reemit_insn_block_notes): Take current block from
3858 nonbind markers. Declare note where it's first set.
3859 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3860 begin stmt markers if enabled.
3861 (notice_source_line): Handle nonbind markers. Fail if their
3862 location is unknown or that of builtins.
3863 (rest_of_handle_final): Convert begin stmt markers to notes if
3864 var-tracking didn't run.
3865 (rest_of_clean_state): Skip begin stmt markers.
3866 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3868 * function.c (allocate_struct_function): Set begin_stmt_markers.
3869 * function.h (struct function): Add debug_marker_count counter
3870 and debug_nonbind_markers flag.
3871 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3872 * gimple-low.c (lower_function_body): Adjust
3873 debug_nonbind_markers.
3874 (lower_stmt): Drop or skip gimple debug stmts.
3875 (lower_try_catch): Skip debug stmts.
3876 * gimple.c (gimple_build_debug_begin_stmt): New.
3877 (gimple_copy): Increment debug_marker_count if copying one.
3878 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3879 * gimplify.c (rexpr_location): New.
3880 (rexpr_has_location): New.
3881 (warn_switch_unreachable_r): Handle gimple debug stmts.
3882 (shortcut_cond_r): Call expr_location.
3884 (find_goto_label): New.
3885 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3887 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3889 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3890 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3891 (LANG_HOOKS_INITIALIZER): ... this.
3892 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3893 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3895 (update_ebb_live_info): Skip debug insn markers.
3896 * lra.c (debug_insn_static_data): Rename to...
3897 (debug_bind_static_data): ... this.
3898 (debug_marker_static_data): New.
3899 (lra_set_insn_recog_data): Select one of the above depending
3901 (lra_update_isn_regno_info): Don't assume debug insns have
3903 (push_insns): Skip debug insns.
3904 * lto-streamer-in.c (input_function): Drop debug stmts
3905 depending on active options. Adjust debug_nonbind_markers.
3906 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3907 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3908 begin stmt marker notes.
3909 (print_insn): Likewise.
3910 * recog.c (extract_insn): Recognize rtl for debug markers.
3911 * rtl.def (DEBUG_MARKER): New.
3912 * tree-inline.c: Include params.h.
3913 (remap_gimple_stmt): Handle nonbind markers.
3914 (maybe_move_debug_stmts_to_successors): Likewise.
3915 (copy_debug_stmt): Likewise.
3916 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
3917 markers regardless of no side effects.
3918 (tsi_link_before): Don't update container's side effects when adding
3919 a begin stmt marker.
3920 (tsi_link_after): Likewise.
3921 (expr_first): Skip begin stmt markers.
3922 (expr_last): Likewise.
3923 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
3924 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3925 Disregard nonbind markers.
3926 * tree.c (make_node_stat): Don't set side effects for begin stmt
3928 (build1_stat): Likewise.
3929 * tree.def (DEBUG_BEGIN_STMT): New.
3930 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
3931 * var-tracking.c (delete_debug_insns): Renamed to...
3932 (delete_vta_debug_insns): ... this.
3933 (reemit_marker_as_note): New.
3934 (vt_initialize): Reemit markers.
3935 (delete_vta_debug_insns): Likewise.
3936 (vt_debug_insns_local): Reemit or delete markers.
3937 (variable_tracking_main_1): Likewise.
3938 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
3939 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
3940 (gimple_debug_nonbind_marker_p): New.
3941 (gimple_build_debug_bind): Adjust.
3942 (gimple_build_debug_begin_stmt): New.
3943 * doc/invoke.texi (max-debug-marker-count): New param.
3944 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
3945 (debug_parameter_ref, debug_marker): New.
3946 (NOTE_INSN_BEGIN_STMT): New.
3947 (DEBUG_INSN): Describe begin stmt markers.
3949 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
3950 without debug insns, we wouldn't, but clean up debug insns
3951 after a control flow insn nevertheless.
3953 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
3954 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
3955 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
3956 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
3957 (rtl_tidy_fallthru_edge): Likewise.
3958 (rtl_verify_fallthru): Likewise.
3959 (rtl_verify_bb_layout): Likewise.
3960 (skip_insns_after_block): Likewise.
3961 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
3962 * dwarf2out.c: Include print-rtl.h.
3963 (dwarf2out_next_real_insn): New.
3964 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
3965 Dump debug binds in asm comments.
3966 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
3967 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
3968 callers to use gsi_start_nondebug_bb instead.
3969 (gsi_after_labels): Skip gimple debug stmts.
3970 (gsi_start_nondebug): New.
3971 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
3972 (proper_loop_form_for_interchange): Adjust.
3973 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
3974 * gimple.h (gimple_seq_last_nondebug_stmt): New.
3975 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
3976 (collect_fallthrough_labels): Likewise.
3977 (should_warn_for_implicit_fallthrough): Likewise.
3978 (warn_implicit_fallthrough_r): Likewise.
3979 (expand_FALLTHROUGH_r): Likewise.
3980 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
3981 (graphite_copy_stmts_from_block): Skip nonbind markers.
3982 * haifa-sched.c (sched_extend_bb): Skip debug insns.
3983 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
3984 * jump.c (clean_barriers): Skip debug insns.
3985 * omp-expand.c (expand_parallel_call): Skip debug insns.
3986 (expand_task_call): Likewise.
3987 (remove_exit_barrier): Likewise.
3988 (expand_omp_taskreg): Likewise.
3989 (expand_omp_for_init_counts): Likewise.
3990 (expand_omp_for_generic): Likewise.
3991 (expand_omp_for_static_nochunk): Likewise.
3992 (expand_omp_for_static_chunk): Likewise.
3993 (expand_omp_simd): Likewise.
3994 (expand_omp_taskloop_for_outer): Likewise.
3995 (expand_omp_taskloop_for_inner): Likewise.
3996 (expand_oacc_for): Likewise.
3997 (expand_omp_sections): Likewise.
3998 (expand_omp_single): Likewise.
3999 (expand_omp_synch): Likewise.
4000 (expand_omp_atomic_load): Likewise.
4001 (expand_omp_atomic_store): Likewise.
4002 (expand_omp_atomic_fetch_op): Likewise.
4003 (expand_omp_atomic_pipeline): Likewise.
4004 (expand_omp_atomic_mutex): Likewise.
4005 (expand_omp_target): Likewise.
4006 (grid_expand_omp_for_loop): Likewise.
4007 (grid_expand_target_grid_body): Likewise.
4008 (build_omp_regions_1): Likewise.
4009 * omp-low.c (check_combined_parallel): Skip debug stmts.
4010 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
4011 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
4012 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
4014 * tree-cfg.c (make_blobs_1): Skip debug stmts.
4015 (make_edges): Likewise.
4016 (cleanup_dead_labels): Likewise.
4017 (gimple_can_merge_blocks_p): Likewise.
4018 (stmt_starts_bb_p): Likewise.
4019 (gimple_block_label): Likewise.
4020 (gimple_redirect_edge_and_branch): Likewise.
4021 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
4023 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
4025 * tree-pretty-print (print_declaration): Omit initializer in slim
4027 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
4029 (eliminate_unnecessary_stmts): Stabilize block removal order.
4030 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
4031 * var-tracking.c (get_first_insn): New.
4032 (vt_emit_notes): Call it.
4033 (vt_initialize): Walk any insns before the first BB.
4034 (delete_debug_insns): Likewise.
4036 * gimple.h (enum gimple_debug_subcode): Add
4037 GIMPLE_DEBUG_BEGIN_STMT.
4038 (gimple_debug_begin_stmt_p): New.
4039 (gimple_debug_nonbind_marker_p): New.
4040 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
4041 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
4042 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
4043 * insn-notes.def (BEGIN_STMT): New.
4044 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
4045 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
4046 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
4047 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
4048 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
4049 (INSN_DEBUG_MARKER_KIND): New.
4050 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
4051 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
4052 (INSN_VAR_LOCATION_PTR): New.
4053 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
4055 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
4056 deps only if debug bind insns are enabled.
4057 (pass_expand::execute): Avoid deep TER and expand
4058 debug locations for debug bind insns only.
4059 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
4060 debug stmts special handling down to debug bind stmts.
4061 * combine.c (try_combine): Narrow debug insns special handling
4062 down to debug bind insns.
4063 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
4064 Narrow debug insns preexisting special handling down to debug
4066 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
4067 handling down to debug bind insns.
4068 * function.c (instantiate_virtual_regs): Skip debug markers,
4069 adjust handling of debug binds.
4070 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
4071 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
4072 * haifa-sched.c (schedule_insn): Narrow special handling of debug
4073 insns to debug bind insns.
4074 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
4075 special handling of debug stmts to debug bind stmts.
4076 * ipa-split.c (split_function): Likewise.
4077 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
4078 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
4080 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
4081 * regrename.c (build_def_use): Likewise.
4082 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
4083 (pass_cprop_hardreg): Narrow special casing of debug insns to
4085 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
4086 * reload1.c (reload): Likewise.
4087 * sese.c (sese_insert_phis_for_liveouts): Narrow special
4088 casing of debug stmts to debug bind stmts.
4089 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
4090 * ssa-iterators.h (num_imm_uses): Likewise.
4091 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
4092 debug stmts to debug bind stmts.
4093 * tree-inline.c (tree_function_versioning): Narrow special casing
4094 of debug stmts to debug bind stmts.
4095 * tree-loop-distribution.c (generate_loops_for_partition):
4096 Narrow special casing of debug stmts to debug bind stmts.
4097 * tree-sra.c (analyze_access_subtree): Narrow special casing
4098 of debug stmts to debug bind stmts.
4099 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
4100 stmts to debug bind stmts.
4101 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
4102 casing of debug stmts to debug bind stmts.
4103 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
4104 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
4105 casing of debug stmts to debug bind stmts.
4106 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
4108 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
4109 debug stmts to debug bind stmts.
4110 (gimple_replace_ssa_lhs): Likewise.
4111 (insert_debug_temp_for_var_def): Likewise.
4112 (insert_debug_temps_for_defs): Likewise.
4113 (reset_debug_uses): Likewise.
4114 * tree-ssanames.c (release_ssa_name_fn): Likewise.
4115 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
4116 (adjust_debug_stmts): Likewise.
4117 (adjust_phi_and_debug_stmts): Likewise.
4118 (vect_do_peeling): Likewise.
4119 * tree-vect-loop.c (vect_transform_loop): Likewise.
4120 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
4121 * var-tracking.c (adjust_mems): Narrow special casing of debug
4122 insns to debug bind insns.
4123 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
4124 (compute_bb_dataflow, vt_find_locations): Likewise.
4125 (vt_expand_loc, emit_notes_for_changes): Likewise.
4126 (vt_init_cfa_base): Likewise.
4127 (vt_emit_notes): Likewise.
4128 (vt_initialize): Likewise.
4129 (vt_finalize): Likewise.
4131 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
4132 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
4133 (next_nonnote_nondebug_insn_bb): New.
4134 (prev_nonnote_nondebug_insn_bb): New.
4135 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
4136 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
4137 (prev_nonnote_nondebug_insn_bb): Declare.
4138 (next_nonnote_nondebug_insn_bb): Declare.
4139 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
4140 * cfgrtl.c (get_last_bb_insn): Likewise.
4141 * lra.c (push_insns): Likewise.
4143 2017-12-11 David Malcolm <dmalcolm@redhat.com>
4146 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
4148 * selftest.c (selftest::test_runner::test_runner): New ctor.
4149 (selftest::test_runner::~test_runner): New dtor.
4150 * selftest.h (class selftest::test_runner): New class.
4152 2017-12-11 Carl Love <cel@us.ibm.com>
4154 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
4155 vec_extract_fp32_from_shortl]): Add #defines.
4156 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
4157 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
4158 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
4159 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
4160 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
4161 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
4162 * doc/extend.texi: Add documentation for the added builtins.
4164 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
4166 PR rtl-optimization/80693
4167 PR rtl-optimization/81019
4168 PR rtl-optimization/81020
4169 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
4170 are not mentioned in i3. Place the REG_UNUSED note on i2,
4171 possibly modified to REG_DEAD, if it did not originate in i3.
4173 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4175 * recog.c (store_data_bypass_p_1): New function.
4176 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
4177 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
4179 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
4181 PR rtl-optimization/83361
4182 * ifcvt.c (if_convert): Call fixup_partitions.
4184 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
4186 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4187 early folding of splat_u{8,16,32}.
4189 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4191 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
4192 output_operand_lossage first argument with capital letter.
4193 (aarch64_override_options): Don't start error and sorry first argument
4194 with capital letter.
4196 2017-12-11 Andi Kleen <ak@linux.intel.com>
4198 PR gcov-profile/83355
4199 * auto-profile.c (string_table::get_index_by_decl): Don't
4200 recurse when abstract origin points to itself.
4202 2017-12-11 Bin Cheng <bin.cheng@arm.com>
4204 PR tree-optimization/83320
4205 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
4206 (prune_datarefs_not_in_loop): Ditto.
4208 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
4210 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
4212 2017-12-10 Jakub Jelinek <jakub@redhat.com>
4214 PR tree-optimization/83337
4215 * gimple-loop-interchange.cc (compute_access_stride): Handle
4216 bitfield DRs properly.
4218 2017-12-09 Jakub Jelinek <jakub@redhat.com>
4220 PR tree-optimization/83338
4221 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
4222 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
4223 vector of unsigned integers to vector of signed integers.
4225 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
4227 PR rtl-optimization/83317
4228 * lra-constraints.c (process_address_1): Add insn code check.
4230 2017-12-08 Michael Matz <matz@suse.de>
4232 Fix PR tree-optimization/83323
4233 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
4234 head-controlled loops and loop BBs.
4235 * common.opt (funroll-and-jam): Remove, instead ...
4236 (floop-unroll-and-jam): ... reuse this option.
4237 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
4238 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
4239 (-floop-unroll-and-jam): ... this option.
4241 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4243 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
4244 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
4245 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
4247 2017-12-08 Richard Biener <rguenther@suse.de>
4250 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
4251 handle accesses outside of zero-sized vars.
4253 2017-12-08 Martin Jambor <mjambor@suse.cz>
4255 PR tree-optimization/83141
4256 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
4257 test for MEM_REFs implicitely changing types with padding. Remove
4259 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
4261 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4263 * config/arc/arc.c (arc_attribute_table): Add exclusions to
4265 * config/avr/avr.c (avr_attribute_table): Likewise.
4266 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
4267 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
4268 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
4269 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
4270 of struct attribute_spec.
4271 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
4273 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
4276 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
4278 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
4281 * profile-count.c (profile_count::from_gcov_type): Move from
4282 profile-count.h; handle overflow.
4283 * profile-count.h (profile_count::from_gcov_type): Move offline.
4285 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
4287 PR rtl-optimization/83304
4288 * combine.c (move_deaths): If we do not know where a register died,
4291 2017-12-08 Richard Biener <rguenther@suse.de>
4293 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
4294 Provide -fopt-info-loop feedback when we interchange in a nest.
4296 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4298 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
4299 for armv6 ARM CPU IDs.
4301 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4303 * common/config/arm/arm-common.c: Include <algorithm>.
4304 (INCLUDE_VECTOR): Define.
4305 (compare_opt_names): New function.
4306 (arm_rewrite_selected_arch): Only strip out extensions that can be
4307 expressed through -mfpu. Sort the remaining extensions
4310 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4312 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
4313 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
4314 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
4315 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
4317 2017-06-08 Tristan Gingold <gindold@adacore.com>
4320 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
4321 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
4323 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4325 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
4326 of struct attribute_spec.
4328 2017-12-08 Julia Koval <julia.koval@intel.com>
4330 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
4331 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
4332 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
4333 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
4334 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
4335 _mm_maskz_dpwssds_epi32): Ditto.
4337 2017-12-08 Richard Biener <rguenther@suse.de>
4339 PR tree-optimization/81303
4340 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
4341 conditions try to create a comparison vector type matching
4342 the data vector type.
4343 (vectorizable_condition): Adjust.
4344 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
4345 Leave invariant conditions alone in case we can vectorize those.
4347 2017-12-08 Julia Koval <julia.koval@intel.com>
4349 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
4350 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
4351 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
4352 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
4353 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
4355 2017-12-08 Julia Koval <julia.koval@intel.com>
4357 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
4358 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
4359 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
4360 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
4361 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
4362 _mm_maskz_dpbusds_epi32): New intrinsics.
4364 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4366 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
4368 (ld<bh>io_signed): Likewise.
4369 (st<bhw>io): Likewise.
4370 * config/nios2/predicates.md (ldstio_memory_operand): Allow
4371 SMALL_INT12 constant integer operand.
4373 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4375 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
4377 (nios2_large_constant_allowed): ...this. Adjust uses.
4378 (nios2_plus_symbolic_constant_p): Rename to...
4379 (nios2_plus_large_constant_p): ...this. Adjust uses.
4380 (nios2_legitimate_address_p): Correct CONST_INT handling.
4381 (nios2_symbolic_memory_operand_p): Rename to...
4382 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
4383 (nios2_large_constant_p): Check for large constant integers too.
4384 (nios2_split_large_constant): Handle constant integers.
4385 (nios2_split_symbolic_memory_operand): Rename to...
4386 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
4387 (nios2_legitimize_constant_address): Handle constant integers.
4388 (r0rel_constant_p): Handle small constant integers.
4389 (nios2_print_operand_address): Handle r0-relative integer addresses.
4390 * config/nios2/nios2-protos.h: Adjust for renamed functions.
4391 * config/nios2/nios2.md: Adjust for renamed functions.
4393 2017-12-07 Andrew Waterman <andrew@sifive.com>
4395 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
4396 (TARGET_HAVE_SRODATA_SECTION): New define.
4397 (riscv_select_section): New function.
4399 2017-12-08 Joseph Myers <joseph@codesourcery.com>
4400 Alexander Monakov <amonakov@ispras.ru>
4401 Jakub Jelinek <jakub@redhat.com>
4404 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
4406 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4408 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
4409 a bool success value. Don't call output_operand_lossage here.
4410 (aarch64_print_ldpstp_address): Return a bool success value.
4411 (aarch64_print_operand_address): Call output_addr_const if
4412 aarch64_print_address_internal fails.
4413 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
4414 'y'; call output_operand_lossage instead. Call output_operand_lossage
4415 if aarch64_print_ldpstp_address fails.
4417 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4419 * tree-vector-builder.h
4420 (tree_vector_builder::binary_encoded_nelts): Declare.
4421 * tree-vector-builder.c
4422 (tree_vector_builder::binary_encoded_nelts): New function.
4423 * fold-const.c (negate_expr_p): Likewise.
4424 (operand_equal_p, fold_checksum_tree): Likewise.
4425 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
4426 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
4427 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
4428 (uniform_vector_p): Likewise.
4429 * varasm.c (const_hash_1, compare_constant): Likewise.
4430 * tree-ssa-ccp.c: Include tree-vector-builder.h.
4431 (valid_lattice_transition): Operate directly on the VECTOR_CST
4433 * ipa-icf.c: Include tree-vector-builder.h.
4434 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
4435 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
4437 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4439 * tree.c (build_vector): Delete.
4440 * tree.h (build_vector): Make static and move into the self-testing
4443 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4445 * vector-builder.h (vector_builder::derived): New const overload.
4446 (vector_builder::elt): New function.
4447 * tree-vector-builder.h (tree_vector_builder::type): New function.
4448 (tree_vector_builder::apply_step): Declare.
4449 * tree-vector-builder.c (tree_vector_builder::apply_step): New
4451 * gimple-fold.h (tree_vector_builder): Declare.
4452 (gimple_build_vector): Take a tree_vector_builder instead of a
4453 type and vector of elements.
4454 * gimple-fold.c (gimple_build_vector): Likewise.
4455 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
4457 (get_initial_defs_for_reduction): Likewise.
4458 (vectorizable_induction): Likewise.
4460 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4462 * tree-vector-builder.h
4463 (tree_vector_builder::new_binary_operation): Declare.
4464 * tree-vector-builder.c
4465 (tree_vector_builder::new_binary_operation): New function.
4466 * fold-const.c (fold_relational_const): Use it.
4467 (const_binop): Likewise. Check that both input vectors have
4468 the same number of elements, thus excluding things like WIDEN_SUM.
4469 Check whether it is possible to operate directly on the encodings
4472 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4474 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
4475 new_unary_operation, operating only on the encoded elements.
4476 (const_unop): Likewise.
4477 (exact_inverse): Likewise.
4478 (distributes_over_addition_p): New function.
4479 (const_binop): Use tree_vector_builder and new_unary_operation
4480 for combinations of VECTOR_CST and INTEGER_CST. Operate only
4481 on the encoded elements unless the encoding is strided and the
4482 operation does not distribute over addition.
4483 (fold_convert_const): Use tree_vector_builder and
4484 new_unary_operation. Operate only on the encoded elements
4485 for truncating integer conversions, or for non-stepped encodings.
4487 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4489 * config/sparc/sparc.c: Include tree-vector-builder.h.
4490 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
4491 * expmed.c: Include tree-vector-builder.h.
4492 (make_tree): Use tree_vector_builder instead of build_vector.
4493 * fold-const.c: Include tree-vector-builder.h.
4494 (const_binop): Use tree_vector_builder instead of build_vector.
4495 (const_unop): Likewise.
4496 (native_interpret_vector): Likewise.
4497 (fold_vec_perm): Likewise.
4498 (fold_ternary_loc): Likewise.
4499 * gimple-fold.c: Include tree-vector-builder.h.
4500 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
4502 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
4503 (simplify_vector_constructor): Use tree_vector_builder instead
4505 * tree-vect-generic.c: Include tree-vector-builder.h.
4506 (add_rshift): Use tree_vector_builder instead of build_vector.
4507 (expand_vector_divmod): Likewise.
4508 (optimize_vector_constructor): Likewise.
4509 * tree-vect-loop.c: Include tree-vector-builder.h.
4510 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
4511 of build_vector. Explicitly use a stepped encoding for
4513 * tree-vect-slp.c: Include tree-vector-builder.h.
4514 (vect_get_constant_vectors): Use tree_vector_builder instead
4516 (vect_transform_slp_perm_load): Likewise.
4517 (vect_schedule_slp_instance): Likewise.
4518 * tree-vect-stmts.c: Include tree-vector-builder.h.
4519 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
4520 (vect_gen_perm_mask_any): Likewise.
4521 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
4522 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
4524 (build_vector_from_val): Likewise. Explicitly use a duplicate
4527 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
4529 * doc/generic.texi (VECTOR_CST): Describe new representation of
4531 * vector-builder.h: New file.
4532 * tree-vector-builder.h: Likewise.
4533 * tree-vector-builder.c: Likewise.
4534 * Makefile.in (OBJS): Add tree-vector-builder.o.
4535 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
4536 * tree-core.h (tree_base): Add a vector_cst field to the u union.
4537 (tree_vector): Change the number of elements to
4538 vector_cst_encoded_nelts.
4539 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
4540 (VECTOR_CST_ELTS): Delete.
4541 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
4542 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
4543 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
4544 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
4545 (VECTOR_CST_ENCODED_ELT): Likewise.
4546 (vector_cst_encoded_nelts): New function.
4547 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4548 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4549 (vector_cst_int_elt, vector_cst_elt): Declare.
4550 * tree.c: Include tree-vector-builder.h.
4551 (tree_code_size): Abort if passed VECTOR_CST.
4552 (tree_size): Update for new VECTOR_CST layout.
4553 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4554 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4555 (build_vector): Use tree_vector_builder.
4556 (vector_cst_int_elt, vector_cst_elt): New functions.
4557 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
4558 encoded elements and then create the vector in the canonical form.
4559 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
4560 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
4561 (tree_c_tests): Call test_vector_cst_patterns.
4562 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
4564 (hash_tree): Likewise.
4565 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
4566 (streamer_write_tree_header): Likewise.
4567 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
4568 (streamer_alloc_tree): Likewise. Update call to make_vector.
4569 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
4571 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4573 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
4574 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
4575 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
4577 2017-12-07 Bin Cheng <bin.cheng@arm.com>
4578 Richard Biener <rguenther@suse.de>
4580 PR tree-optimization/81303
4581 * Makefile.in (gimple-loop-interchange.o): New object file.
4582 * common.opt (floop-interchange): Reuse the option from graphite.
4583 * doc/invoke.texi (-floop-interchange): Ditto. New document for
4584 -floop-interchange and mention it for -O3.
4585 * opts.c (default_options_table): Enable -floop-interchange at -O3.
4586 * gimple-loop-interchange.cc: New file.
4587 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
4588 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
4589 * passes.def (pass_linterchange): New pass.
4590 * timevar.def (TV_LINTERCHANGE): New time var.
4591 * tree-pass.h (make_pass_linterchange): New declaration.
4592 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
4593 interchange. Record IV before/after increment in new parameters.
4594 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
4595 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
4597 (check_reduction_path): ...New function here.
4598 * tree-vectorizer.h (check_reduction_path): New declaration.
4600 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
4603 PR rtl-optimization/80818
4604 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
4605 always early clobbered.
4606 * lra-lives.c (process_bb_lives): Check input hard regs for early
4607 clobbered non-operand hard reg.
4609 2017-12-07 Jakub Jelinek <jakub@redhat.com>
4612 * tree-cfg.c (verify_gimple_assign_binary): Don't require
4613 types_compatible_p, just that TYPE_MODE is the same.
4615 2017-12-07 Martin Sebor <msebor@redhat.com>
4618 * attribs.c (empty_attribute_table): Initialize new member of
4619 struct attribute_spec.
4620 (decl_attributes): Add argument. Handle mutually exclusive
4621 combinations of attributes.
4622 (selftests::test_attribute_exclusions): New function.
4623 (selftests::attribute_c_tests): Ditto.
4624 * attribs.h (decl_attributes): Add default argument.
4625 * selftest.h (attribute_c_tests): Declare.
4626 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
4627 * tree-core.h (attribute_spec::exclusions, exclude): New type and
4629 * doc/extend.texi (Common Function Attributes): Update const and pure.
4630 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
4631 of struct attribute_spec.
4632 * config/arc/arc.c (arc_attribute_table): Same.
4633 * config/arm/arm.c (arm_attribute_table): Same.
4634 * config/avr/avr.c ( avr_attribute_table): Same.
4635 * config/bfin/bfin.c (bfin_attribute_table): Same.
4636 * config/cr16/cr16.c (cr16_attribute_table): Same.
4637 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
4638 * config/h8300/h8300.c (h8300_attribute_table): Same.
4639 * config/i386/i386.c (ix86_attribute_table): Same.
4640 * config/ia64/ia64.c (ia64_attribute_table): Same.
4641 * config/m32c/m32c.c (m32c_attribute_table): Same.
4642 * config/m32r/m32r.c (m32r_attribute_table): Same.
4643 * config/m68k/m68k.c (m68k_attribute_table): Same.
4644 * config/mcore/mcore.c (mcore_attribute_table): Same.
4645 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
4646 * config/mips/mips.c (mips_attribute_table): Same.
4647 * config/msp430/msp430.c (msp430_attribute_table): Same.
4648 * config/nds32/nds32.c (nds32_attribute_table): Same.
4649 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
4650 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
4651 * config/rl78/rl78.c (rl78__attribute_table): Same.
4652 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
4653 * onfig/rx/rx.c (rx_attribute_table): Same.
4654 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
4655 * config/sh/sh.c (sh_attribute_table): Same.
4656 * config/sparc/sparc.c (sparc_attribute_table): Same.
4657 * config/spu/spu.c (spu_attribute_table): Same.
4658 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
4659 * config/v850/v850.c (v850_attribute_table): Same.
4660 * config/visium/visium.c (visium_attribute_table): Same.
4662 2017-12-07 Tamar Christina <tamar.christina@arm.com>
4665 * config/arm/arm.c (INCLUDE_STRING): Define.
4666 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
4667 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
4670 2017-12-07 Michael Matz <matz@suse.de>
4672 Add unroll and jam pass
4674 * gimple-loop-jam.c: New file.
4675 * Makefile.in (OBJS): Add gimple-loop-jam.o.
4676 * common.opt (funroll-and-jam): New option.
4677 * opts.c (default_options_table): Add unroll-and-jam at -O3.
4678 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
4679 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
4680 * passes.def: Add pass_loop_jam.
4681 * timevar.def (TV_LOOP_JAM): Add.
4682 * tree-pass.h (make_pass_loop_jam): Declare.
4683 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
4684 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
4685 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
4686 to flow_loop_tree_node_add.
4687 (duplicate_subloops, copy_loops_to): Append to sibling list.
4688 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
4689 * doc/invoke.texi (-funroll-and-jam): Document new option.
4690 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
4692 2017-12-07 Richard Biener <rguenther@suse.de>
4694 PR tree-optimization/83296
4695 PR tree-optimization/67769
4696 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
4697 flow sensitive info in an unrelated BB.
4698 (value_replacement): Use reset_flow_sensitive_info.
4699 (minmax_replacement): Reset flow sensitive info on the def
4700 we move. Do not reset flow sensitive info in the whole BB
4701 we move the stmt to.
4702 (abs_replacement): Likewise.
4704 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4707 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4708 rs6000_cpu to the given -mcpu=, or to the default processor.
4710 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4712 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
4713 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
4714 instead of rs6000_cpu_attr.
4715 (rs6000_adjust_cost): Ditto.
4716 (is_microcoded_insn): Ditto.
4717 (rs6000_adjust_priority): Ditto.
4718 (rs6000_issue_rate): Ditto.
4719 (rs6000_use_sched_lookahead): Ditto.
4720 (rs6000_use_sched_lookahead_guard): Ditto.
4721 (rs6000_sched_reorder): Ditto.
4722 (force_new_group): Ditto.
4723 * config/rs6000/rs6000.md (cpu attribute): Ditto.
4724 (group_ending_nop): Ditto.
4726 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4728 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
4729 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
4730 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
4731 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
4733 (rs6000_reassociation_width): Ditto.
4734 (rs6000_emit_epilogue): Ditto.
4735 (rs6000_adjust_cost): Ditto.
4736 (is_microcoded_insn): Ditto.
4737 (is_cracked_insn): Ditto.
4738 (rs6000_adjust_priority): Ditto.
4739 (rs6000_sched_reorder): Ditto.
4740 (rs6000_sched_reorder2): Ditto.
4741 (insn_must_be_first_in_group): Ditto.
4742 (insn_must_be_last_in_group): Ditto.
4743 (rs6000_register_move_cost): Ditto.
4744 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
4747 2017-12-07 Julia Koval <julia.koval@intel.com>
4749 * config.gcc: Add vaesintrin.h.
4750 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
4752 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
4753 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
4755 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4756 * config/i386/immintrin.h: Include vaesintrin.h.
4757 * config/i386/sse.md (vaesdec_<mode>): New pattern.
4758 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
4759 _mm_aesdec_epi128): New intrinsics.
4761 2017-12-06 David Malcolm <dmalcolm@redhat.com>
4763 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
4764 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
4765 c-family/c-spellcheck.cc.
4766 (best_macro_match::best_macro_match): Likewise.
4768 (struct edit_distance_traits<cpp_hashnode *>): Move to
4769 c-family/c-spellcheck.h.
4770 (class best_macro_match): Likewise.
4772 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4774 PR tree-optimization/83293
4775 * gimple-ssa-strength-reduction.c (insert_initializers): Use
4776 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
4777 might insert into empty bb.
4780 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
4781 simplify for plus with :c added, and pointer_plus without that.
4782 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
4783 with undefined overflow and the conversion is not widening,
4784 perform negation in utype and only convert to type afterwards.
4785 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
4786 simplify for plus with :c added, and pointer_plus without that.
4787 If type is integral with undefined overflow and the conversion is
4788 not widening, perform minus in utype and only convert to type
4789 afterwards. Move the last pointer_diff_expr simplify into the
4792 2017-12-06 Martin Sebor <msebor@redhat.com>
4794 PR tree-optimization/82646
4795 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
4796 strncpy, not maxlen.
4798 2017-12-06 Martin Sebor <msebor@redhat.com>
4800 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
4803 PR tree-optimization/83075
4804 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
4805 strncat/strncpy don't change length of source string.
4807 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4810 2017-11-29 Martin Aberg <maberg@gaisler.com>
4812 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4813 to prevent b2bst errata sequence.
4814 (sqrtdf2_fix): Likewise.
4816 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4818 PR tree-optimization/81945
4819 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4820 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4821 to dest_cfun has orig_loop_num set, either remap it to the new
4822 loop number if the loop got moved too, or clear it.
4824 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4826 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4828 (thunderx2t99_nothing): New insn reservation.
4829 (thunderx2t99_mrs): New insn reservation.
4830 (thunderx2t99_multiple): New insn reservation.
4831 (thunderx2t99_alu_basi): Add bfx to reservation.
4832 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4834 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4837 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4839 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4841 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4842 tree-ssa-pre.c::remove_dead_inserted_code.
4843 * tree-ssa-dce.h: New file.
4844 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
4845 (remove_dead_inserted_code): Move and rename to function
4846 tree-ssa-dce.c::simple_dce_from_worklist.
4847 (pass_pre::execute): Update use.
4849 2017-12-05 Richard Biener <rguenther@suse.de>
4851 PR tree-optimization/83277
4852 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4853 to code-gen liveout vars.
4855 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4857 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4858 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4859 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4862 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4864 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4866 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4867 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4870 2017-12-05 Richard Biener <rguenther@suse.de>
4872 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4873 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4874 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4875 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4877 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4879 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4881 2017-12-05 Martin Liska <mliska@suse.cz>
4882 Jakub Jelinek <jakub@redhat.com>
4884 * doc/invoke.texi: Document the options.
4885 * flag-types.h (enum sanitize_code): Add
4886 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4887 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4888 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4889 * opts.c: Define new sanitizer options.
4890 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4891 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4893 2017-12-05 Julia Koval <julia.koval@intel.com>
4895 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4896 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4897 (ix86_handle_option): Handle -mavx512vnni.
4898 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4899 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4900 * config/i386/i386-c (__AVX512VNNI__): New.
4901 * config/i386/i386.c (ix86_target_string): Handle new option.
4902 (ix86_valid_target_attribute_inner_p): Handle new option.
4903 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4904 * config/i386/i386.opt (mavx512vnni): New option.
4906 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4909 * athlon.md: Disable for generic.
4910 * haswell.md: Enable for generic.
4911 * i386.c (ix86_sched_init_global): Add core hooks for generic.
4912 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
4914 (ix86_adjust_cost): Move generic to haswell path.
4916 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
4918 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
4919 instead of MEM_P in a couple more places. Fix formatting issues.
4921 2017-12-04 Jim Wilson <jimw@sifive.com>
4923 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
4924 instead of GP_REG_LAST-1.
4925 (riscv_adjust_libcall_cfi_prologue): Likewise.
4926 (riscv_adjust_libcall_cri_epilogue): Likewise.
4927 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
4930 2017-12-04 Luis Machado <luis.machado@linaro.org>
4932 * ipa-pure-const.c (check_decl): Add missing newline.
4933 (state_from_flags): Likewise.
4935 2017-12-04 Jeff Law <law@redhat.com>
4937 PR tree-optimizatin/78496
4938 * gimple-ssa-evrp-analyze.h
4939 (evrp_range_analyzer::get_vr_values): Simplify.
4940 * gimple-ssa-evrp-analyze.c: Corresponding changes.
4941 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
4942 and gimple-ssa-evrp-analyze.h.
4943 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
4944 (simplify_stmt_for_jump_threading): Copy a blob of code from
4945 tree-vrp.c to use ranges to simplify statements.
4946 (dom_opt_dom_walker::before_dom_children): Call
4947 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
4948 (dom_opt_dom_walker::after_dom_children): Similarly for
4949 evrp_range_analyzer::leave.
4950 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
4953 * gimple-ssa-evrp-analyze.c
4954 (evrp_range_analyzer::extract_range_from_stmt): Always use
4955 vr_values::update_value_range so preexisting range info is
4956 medged with new range info, even if the new range is VR_VARYING.
4958 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4960 * combine.c: Adjust comment.
4961 (use_crosses_set_p): Delete.
4962 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
4963 (try_combine): Ditto.
4965 2017-12-04 Richard Biener <rguenther@suse.de>
4967 PR tree-optimization/83255
4968 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
4969 Re-add zero-iteration check.
4971 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4973 PR rtl-optimization/83245
4974 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
4975 hard registers as earlyclobber, also if not in an asm.
4977 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4981 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4984 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4985 rs6000_cpu based on cpu_index, not tune_index.
4987 2017-12-04 Richard Biener <rguenther@suse.de>
4989 PR tree-optimization/83238
4990 * graphite-scop-detection.c (scop_detection::merge_sese): Make
4991 code match comment, rejecting invalid SESE regions.
4993 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
4995 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
4996 require base operand is a REG_POINTER prior to reload on targets
4997 with non-equivalent space registers.
4999 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
5001 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
5002 (update_specialized_profile): Fix updating of counts.
5003 (perhaps_add_new_callers): Likewise.
5005 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
5008 * x86-tune.def: Remove obsolette FIXMEs.
5009 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
5010 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
5011 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
5013 (X86_TUNE_PAD_RETURNS): Disable for generic.
5015 2017-12-02 Jakub Jelinek <jakub@redhat.com>
5017 PR tree-optimization/83170
5018 PR tree-optimization/83241
5019 * gimple-ssa-store-merging.c
5020 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
5021 gimple_vuse (ins_stmt) in case it has changed.
5022 (imm_store_chain_info::output_merged_store): Likewise.
5024 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
5028 * tree-cfg.c (pass_warn_function_return::execute): Handle
5029 __builtin_ubsan_handle_missing_return like __builtin_unreachable
5030 with BUILTINS_LOCATION.
5034 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
5035 is BLKmode for vector field with vector raw mode, use TYPE_MODE
5036 instead of DECL_MODE.
5038 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
5039 last argument to rtx pointer.
5040 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
5041 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
5042 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
5043 depending on the chosen ISAs.
5044 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
5045 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
5046 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
5048 * config/i386/sse.md (mov<mode>_internal): Likewise.
5049 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
5051 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5053 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
5055 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5057 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
5058 parameters from prototype.
5059 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
5060 parameters. Don't print a comment.
5061 (emit_fusion_gpr_load): Adjust.
5062 (emit_fusion_load_store): Adjust.
5063 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
5064 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
5065 comment on the second line.
5067 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5070 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
5071 rs6000_cpu based on cpu_index, not tune_index.
5073 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5075 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
5076 which_alternative instead of which_alternative + 1.
5077 (output_asm_insn): Print an extra tab if the template is short.
5079 2017-12-01 Jim Wilson <jimw@sifive.com>
5081 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
5083 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
5084 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
5086 * doc/tm.texi: Regenerate.
5088 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
5091 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
5092 whether we can allocate pseudos before trying to fix an address.
5093 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
5094 memory address is indexed or indirect.
5095 (floatuns_<mode>si2_hw2): Likewise.
5097 2017-12-01 Jason Merrill <jason@redhat.com>
5099 * Makefile.in (TAGS): Add c-family/*.cc.
5101 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
5103 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
5104 (call_value_insn): Likewise.
5105 (sibcall_insn): Likewise.
5106 (sibcall_value_insn): Likewise.
5107 (movsi_aarch64): Likewise.
5108 (movdi_aarch64): Likewise.
5109 (add_losym_): Likewise.
5110 (ldr_got_small_): Likewise.
5111 (ldr_got_small_sidi): Likewise.
5112 (ldr_got_small_28k_): Likewise.
5113 (ldr_got_small_28k_sidi): Likewise.
5114 * config/aarch64/aarch64.c (aarch64_print_address_internal):
5115 Move output_addr_const to symbolic case. Add error check.
5117 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5119 * config/s390/predicates.md (plus16_Q_operand): New predicate.
5120 * config/s390/s390.md: Disable MVC merging peephole if it would
5121 disable operand forwarding.
5122 (new peephole2): Split MVCs if it would turn them into up to 2
5125 2017-12-01 Richard Biener <rguenther@suse.de>
5127 PR tree-optimization/83232
5128 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
5129 detection of same access. Instead of breaking the group here
5130 do not consider the duplicate. Add comment explaining real fix.
5132 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
5134 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
5136 2017-12-01 Sudakshina Das <sudi.das@arm.com>
5138 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
5140 2017-12-01 Jakub Jelinek <jakub@redhat.com>
5142 * function.h (struct function): Remove cilk_frame_decl,
5143 is_cilk_function and calls_cilk_spawn fields.
5144 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
5146 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
5147 cilk_elemental field.
5148 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
5149 * target.def: Adjust comment.
5150 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
5151 Don't test cilk_elemental.
5153 PR tree-optimization/83233
5154 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
5155 bswap_stat name for the struct.
5158 * tree.h (SWITCH_BREAK_LABEL_P): Define.
5159 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
5160 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
5161 SWITCH_BREAK_LABEL_P set on the label.
5162 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
5163 added for default case if it was missing and not all cases covered.
5164 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
5165 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
5167 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
5168 to avoid -Wimplicit-fallthrough warning.
5169 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
5170 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
5172 PR tree-optimization/83221
5173 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
5175 (init_reassoc): Formatting fix.
5178 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
5179 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
5183 * tree-cfg.c: Include asan.h.
5184 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
5186 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5188 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
5190 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5192 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
5193 vpcmpeqd instruction.
5195 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5197 * config/i386/i386.c (standard_sse_constant_opcode): Fix
5198 registers type for 128bit mode.
5200 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5202 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
5204 * gengtype-state.c (read_state_pair): Likewise.
5205 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
5206 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
5207 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
5208 * ifcvt.c (noce_try_store_flag_constants): Likewise.
5209 * tree-ssa-ccp.c (ccp_finalize): Likewise.
5210 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
5211 * builtins.c (fold_builtin_3): Likewise.
5212 * graphite-scop-detection.c
5213 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
5214 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
5217 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
5220 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
5221 parameter. Return true if ignore_decl_rtl_set_p is true and other
5222 conditions are satisfied.
5223 * asan.h (asan_protect_global): Add new parameter.
5224 * varasm.c (categorize_decl_for_section): Pass true as second parameter
5225 to asan_protect_global calls.
5227 2017-11-30 Jim Wilson <jimw@sifive.com>
5229 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
5230 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
5231 -mstrict-align, add info on default value. Delete redundant lines for
5232 -mabi. Add missing -mexplicit-relocs docs.
5234 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5236 * config/arc/arc.md (trap): New pattern.
5238 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5240 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
5241 instruction to end into a delay slot.
5242 * config/arc/arc.md (cond_delay_insn): Check if the instruction
5243 can be placed into a delay slot against reg_note.
5245 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5247 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
5248 labels number of usages.
5250 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5252 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
5254 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
5256 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5259 * internal-fn.c (expand_mul_overflow): Optimize unsigned
5260 multiplication by power of 2 constant into two shifts + comparison.
5262 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
5265 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
5267 2017-11-30 Richard Biener <rguenther@suse.de>
5269 PR tree-optimization/83202
5270 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
5271 allow_peel argument and guard peeling.
5272 (canonicalize_loop_induction_variables): Likewise.
5273 (canonicalize_induction_variables): Pass false.
5274 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
5275 peeling from cunrolli.
5277 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5279 * combine.c (try_combine): Print a message to dump file whenever
5280 I0, I1, or I2 cannot be combined into I3.
5282 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5284 PR rtl-optimization/83156
5285 PR rtl-optimization/82621
5286 * combine.c (try_combine): Don't split an I2 if one of the dests is
5287 set again before I3. Allow unused dests.
5289 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5291 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
5293 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
5295 PR rtl-optimization/80818
5296 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
5297 recursively. Use insn code for clobber.
5298 (lra_set_insn_recog_data): Pass the new arg to
5299 collect_non_operand_hard_regs.
5300 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
5302 (lra_update_insn_regno_info): Pass insn to
5303 add_regs_to_insn_regno_info.
5305 2017-11-29 Jim Wilson <jimw@sifive.com>
5306 Andrew Waterman <andrew@sifive.com>
5308 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
5309 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
5310 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
5311 use SINGLE_SHIFT_COST.
5312 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
5313 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
5315 2017-11-29 Julia Koval <julia.koval@intel.com>
5317 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
5318 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
5319 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
5320 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
5321 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
5322 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
5323 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
5324 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
5325 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
5326 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
5327 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
5328 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
5329 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
5330 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
5331 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
5332 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
5333 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
5334 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
5335 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
5336 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
5337 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
5338 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
5339 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
5340 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
5341 __builtin_ia32_vpshldv_v2di_mask,
5342 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
5343 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
5344 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
5346 2017-11-29 Julia Koval <julia.koval@intel.com>
5348 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
5349 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
5350 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
5351 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
5352 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
5353 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
5354 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
5355 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
5356 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
5357 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
5358 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
5359 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
5360 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
5361 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
5362 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
5363 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
5364 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
5365 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
5366 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5367 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
5368 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
5370 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5372 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
5373 movsi_pic_gotdata_op instruction as a load for the UT699 errata
5376 2017-11-29 Martin Aberg <maberg@gaisler.com>
5378 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
5379 to prevent b2bst errata sequence.
5380 (sqrtdf2_fix): Likewise.
5382 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5384 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
5385 (div_sqrt_insn_p): New function.
5386 (sparc_do_work_around_errata): Insert NOP instructions to
5387 prevent sequences that could trigger the TN-0013 errata for
5388 certain LEON3 processors.
5389 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
5390 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
5391 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
5392 (in_branch_delay): Prevent div and sqrt in delay slot if
5394 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
5396 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5398 * config/sparc/sparc.c (atomic_insn_p): New function.
5399 (sparc_do_work_around_errata): Insert NOP instructions to
5400 prevent sequences that could trigger the TN-0010 errata for
5402 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
5403 instruction referable in atomic_insns_p.
5405 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5407 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
5408 (atomic_compare_and_swap_leon3_1): Likewise.
5411 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5413 * config/sparc/sparc.c (fpop_insn_p): New function.
5414 (sparc_do_work_around_errata): Insert NOP instructions to
5415 prevent sequences that could trigger the TN-0012 errata for
5417 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
5418 * config/sparc/sparc.md (fix_gr712rc): New attribute.
5419 (in_branch_annul_delay): Prevent floating-point instructions
5420 in delay slot of annulled integer branch.
5422 2017-11-29 Richard Biener <rguenther@suse.de>
5424 PR tree-optimization/83202
5425 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
5427 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
5428 nodes vectorized to the same stmts multiple times.
5429 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
5430 (vect_analyze_slp_instance): Adjust.
5431 (scalar_stmts_to_slp_tree_map_t): New typedef.
5432 (vect_schedule_slp_instance): Add a map recording the SLP node
5433 representing the vectorized stmts for a set of scalar stmts.
5434 Avoid code-generating redundancies.
5435 (vect_schedule_slp): Allocate map and pass it down.
5437 2017-11-29 Nathan Sidwell <nathan@acm.org>
5440 * tree.c (build_complex_type): Fix canonicalization. Only fill in
5443 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
5445 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
5446 cases for printing LDP/STP memory addresses.
5447 (aarch64_print_address_internal): Renamed from
5448 aarch64_print_operand_address, added parameter, add Pmode check.
5449 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
5450 (aarch64_print_operand_address): Indirect to
5451 aarch64_print_address_internal.
5452 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
5453 'y' operand output specifier.
5455 2017-11-29 Jakub Jelinek <jakub@redhat.com>
5458 * tree.c (build_simple_mem_ref_loc): Handle
5459 get_addr_base_and_unit_offset returning a MEM_REF.
5462 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
5465 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
5468 2017-11-28 David Malcolm <dmalcolm@redhat.com>
5470 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
5472 (layout::move_to_column): Likewise.
5474 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5477 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
5478 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
5480 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5481 Martin Jambor <mjambor@suse.cz>
5484 * tree.h (expr_type_first_operand_type_p): Declare
5485 * tree.c (expr_type_first_operand_type_p): New function.
5486 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
5487 (ipa_value_from_jfunc): Adjust declaration.
5488 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
5489 Use it as result type for arithmetics, unless it is NULL in which case
5490 be more conservative.
5491 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
5492 ipa_get_jf_pass_through_result.
5493 (propagate_vals_across_pass_through): Likewise.
5494 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
5495 is to propagate_vals_across_pass_through.
5496 (propagate_constants_across_call): Pass PARM_TYPE to
5497 propagate_scalar_across_jump_function.
5498 (find_more_scalar_values_for_callers_subset): Pass parameter type to
5499 ipa_value_from_jfunc.
5500 (cgraph_edge_brings_all_scalars_for_node): Likewise.
5501 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
5502 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
5503 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
5504 target_type, pass it to ipa_value_from_jfunc.
5505 (update_indirect_edges_after_inlining): Pass parameter type to
5506 try_make_edge_direct_simple_call.
5508 2017-11-28 Jeff Law <law@redhat.com>
5510 * gimple-ssa-evrp-analyze.c
5511 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
5512 refine ranges if scev_initialized_p returns true.
5513 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
5515 2017-11-28 Julia Koval <julia.koval@intel.com>
5517 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
5518 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
5519 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
5520 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
5521 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
5522 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
5523 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
5524 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
5525 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
5526 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
5527 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
5528 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
5529 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
5530 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
5531 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
5532 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
5533 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
5534 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
5535 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
5536 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
5537 __builtin_ia32_vpshrd_v2di_mask): New builtins.
5538 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
5540 2017-11-28 Julia Koval <julia.koval@intel.com>
5542 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
5543 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
5544 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
5545 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
5546 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
5547 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
5548 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
5549 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
5550 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
5551 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
5552 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
5553 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
5554 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
5555 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
5556 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
5557 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
5558 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
5559 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
5560 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
5561 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
5562 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
5563 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
5564 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
5565 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
5566 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
5567 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
5568 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
5569 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
5570 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
5571 __builtin_ia32_vpshld_v2di_mask): New builtins.
5572 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5573 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
5575 2017-11-28 Richard Biener <rguenther@suse.de>
5577 PR tree-optimization/80776
5578 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
5580 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
5582 (evrp_range_analyzer::record_ranges_from_incoming_edges):
5583 If the incoming edge is an effective fallthru because the other
5584 edge only reaches a __builtin_unreachable () then record ranges
5585 derived from the controlling condition in SSA info.
5586 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
5587 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5589 2017-11-28 Olivier Hainque <hainque@adacore.com>
5591 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
5592 on mingw build hosts.
5594 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5597 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
5599 2017-11-28 Richard Biener <rguenther@suse.de>
5602 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
5603 copies generated from memcpy use a character array as reference
5606 2017-11-28 Julia Koval <julia.koval@intel.com>
5607 Sebastian Peryt <sebastian.peryt@intel.com>
5609 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
5610 c-family/c-cilkplus.o, c-family/array-notation-common.o,
5611 cilk-common.o, cilk.h, cilk-common.c): Remove.
5613 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
5614 * builtins.c (is_builtin_name): Remove cilkplus condition.
5615 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
5616 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
5617 cilk-builtins.def, cilkplus.def): Remove.
5618 * cif-code.def (CILK_SPAWN): Remove.
5619 * cilk-builtins.def: Delete.
5620 * cilk-common.c: Ditto.
5622 * cilkplus.def: Ditto.
5623 * config/darwin.h (fcilkplus): Delete.
5624 * cppbuiltin.c: Ditto.
5625 * doc/extend.texi: Remove cilkplus doc.
5626 * doc/generic.texi: Ditto.
5627 * doc/invoke.texi: Ditto.
5628 * doc/passes.texi: Ditto.
5629 * gcc.c (fcilkplus): Remove.
5630 * gengtype.c (cilk.h): Remove.
5631 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
5633 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
5635 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
5637 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
5638 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
5639 cilkplus conditions.
5640 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
5641 inline_read_section): Ditto.
5642 * ipa-inline-analysis.c (cilk.h): Remove.
5643 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
5644 * lto-wrapper.c (merge_and_complain, append_compiler_options,
5645 append_linker_options): Remove condition for fcilkplus.
5646 * lto/lto-lang.c (cilk.h): Remove.
5647 (lto_init): Remove condition for fcilkplus.
5648 * omp-expand.c (expand_cilk_for_call): Delete.
5649 (expand_omp_taskreg, expand_omp_for_static_chunk,
5650 expand_omp_for): Remove cilkplus
5652 (expand_cilk_for): Delete.
5653 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
5654 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
5655 execute_lower_omp, diagnose_sb_0): Ditto.
5656 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
5657 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5658 * tree-nested.c: Ditto.
5659 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
5660 (dump_generic_node): Ditto.
5661 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5662 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
5664 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
5666 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5668 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
5669 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
5670 for vector mode and !TARGET_SIMD.
5672 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5674 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
5676 * tree.h (SWITCH_LABELS): Remove.
5677 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
5678 assert SWITCH_BODY is non-NULL.
5679 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
5681 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
5683 PR tree-optimization/80788
5684 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
5685 has TREE_OVERFLOW set, call drop_tree_overflow.
5687 2017-11-28 Richard Biener <rguenther@suse.de>
5689 PR tree-optimization/83158
5690 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
5692 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
5695 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
5696 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
5698 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
5700 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
5701 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
5703 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5706 * varasm.c (compare_constant): Take the mode of the constants into
5707 account when comparing floating point constants.
5709 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
5711 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
5712 from explicit instantiation of debug_helper.
5713 * vec.h (DEFINE_DEBUG_VEC): Ditto.
5715 2017-11-27 Richard Biener <rguenther@suse.de>
5717 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
5720 2017-11-27 Richard Biener <rguenther@suse.de>
5722 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
5723 (type_hash_canon): Also clear int_cst_hash_table entry for
5725 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
5727 2017-11-27 Tamar Christina <tamar.christina@arm.com>
5729 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5730 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5731 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5733 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5735 * hash-map.h (gt_cleare_cache): Avoid UB.
5737 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
5739 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
5740 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
5741 (decide_unroll_runtime_iterations): Likewise.
5742 (decide_unroll_stupid): Likewise.
5744 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5747 * config/i386/i386.md: Add a loop with incssp.
5749 2017-11-27 Martin Jambor <mjambor@suse.cz>
5751 PR tree-optimization/81248
5752 * tree-sra.c (splice_param_accesses): Remove size check.
5753 (decide_one_param_reduction): Fix size check.
5754 * gimple-pretty-print.c (dump_profile): Silence warning.
5755 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
5757 2017-11-27 Jakub Jelinek <jakub@redhat.com>
5760 * dbxout.c (lastlineno): New variable.
5761 (dbx_debug_hooks): Use dbxout_switch_text_section as
5762 switch_text_section debug hook.
5763 (dbxout_function_end): Switch to current_function_section
5764 rather than function_section. If crtl->has_bb_partition,
5765 output just one N_FUN, depending on in_cold_section_p.
5766 (dbxout_source_line): Remember last lineno in lastlineno.
5767 (dbxout_switch_text_section): New function.
5768 (dbxout_function_decl): Adjust dbxout_block caller.
5769 (dbx_block_with_cold_children): New function.
5770 (dbxout_block): Return true if any LBRAC/RBRAC have been
5771 emitted. Use dbx_block_with_cold_children at depth == 0
5772 in second partition. Add PARENT_BLOCKNUM argument, pass
5773 it optionally adjusted to children. Output LBRAC/RBRAC
5774 around recursive call only if the block is in the current
5775 partition, if not and anything was output, emit empty
5777 * final.c (final_scan_insn): Compute cold_function_name
5778 before calling switch_text_section debug hook. Call
5779 that hook even if dwarf2out_do_frame if not emitting
5783 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
5784 TREE_READONLY decls.
5786 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5788 PR rtl-optimization/82488
5789 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
5791 2017-11-26 Julia Koval <julia.koval@intel.com>
5793 * config/i386/i386.c (processor_target_table): Add skylake_cost for
5795 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
5798 2017-11-26 Julia Koval <julia.koval@intel.com>
5800 * config/i386/driver-i386.c (host_detect_local_cpu):
5801 Detect skylake-avx512.
5803 2017-11-26 Julia Koval <julia.koval@intel.com>
5805 * config.gcc: Add -march=cannonlake.
5806 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5807 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5808 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5809 (PTA_CANNONLAKE): New.
5810 (processor_target_table): Add cannonlake.
5811 (ix86_option_override_internal): Ditto.
5812 (fold_builtin_cpu): Ditto.
5813 (get_builtin_code_for_version): Handle cannonlake.
5814 (M_INTEL_COREI7_CANNONLAKE): New.
5815 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5816 * doc/invoke.texi: Add -march=cannonlake.
5818 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5820 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5821 (try_init_one_plugin): Alternative implementation for MinGW.
5822 * Makefile.in (plugin_implib): New.
5823 (gengtype-lex.c): Fix broken AIX workaround.
5824 * configure: Regenerate.
5825 * doc/plugins.texi: Document support for MinGW.
5827 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5829 PR rtl-optimization/81553
5830 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5831 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5832 is a shift where C1 has different mode than the whole shift, use C1's
5833 mode for MULT rather than the shift's mode.
5836 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5837 builtins not enabled in the currently selected ISA.
5839 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5841 PR tree-optimization/71026
5842 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5843 (insert_reciprocals): Change to insert reciprocals before a division
5844 by a square and to insert the square of a reciprocal.
5845 (execute_cse_reciprocals_1): Change to consider division by a square.
5846 (register_division_in): Add importance parameter.
5848 2017-11-24 Richard Biener <rguenther@suse.de>
5850 PR tree-optimization/82402
5851 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5852 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5854 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5856 * match.pd (0-ptr): New transformation.
5858 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5861 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5862 when skipping checking.
5864 2017-11-24 Richard Biener <rguenther@suse.de>
5866 PR tree-optimization/83128
5867 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5868 (vn_reference_lookup_3): Likewise.
5870 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5873 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5874 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5877 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5878 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5879 check for the latter with gimple_call_builtin_p. Do not handle
5880 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5882 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5884 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5887 2017-11-23 Julia Koval <julia.koval@intel.com>
5889 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5890 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5891 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5892 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5893 _mm512_maskz_expandloadu_epi16): New intrinsics.
5894 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5895 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5896 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5897 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5898 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5899 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5900 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5901 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5902 _mm256_maskz_expandloadu_epi8): New intrinsics.
5903 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5904 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5905 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5906 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5907 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5908 * config/i386/sse.md (VI248_VLBW): New iterator.
5909 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5911 2017-11-23 Julia Koval <julia.koval@intel.com>
5913 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
5914 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
5915 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
5916 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
5917 _mm512_mask_compressstoreu_epi16): New.
5918 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
5919 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
5920 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
5921 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
5922 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
5923 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
5924 _mm256_mask_compressstoreu_epi8): New.
5925 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
5926 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
5927 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
5928 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
5929 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
5930 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
5931 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
5932 __builtin_ia32_compresshi128_mask,
5933 __builtin_ia32_compressstoreuqi512_mask,
5934 __builtin_ia32_compressstoreuhi512_mask,
5935 __builtin_ia32_compressstoreuqi256_mask,
5936 __builtin_ia32_compressstoreuqi128_mask,
5937 __builtin_ia32_compressstoreuhi256_mask,
5938 __builtin_ia32_compressstoreuhi128_mask): New builtins.
5939 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
5941 (ix86_expand_special_args_builtin): Handle new types.
5942 (s4fma_expand): Handle new builtin array.
5943 * config/i386/immintrin.h: Include new headers.
5944 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
5945 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
5947 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5949 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
5950 fixes. Declare temp and g variables at the top in order to avoid
5951 {} in most of the cases.
5953 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
5955 * match.pd (ptr-0): New transformation.
5957 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5959 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
5960 (struct addr_mode_cost_table): New.
5961 (struct tune_params): Add field addr_mode_costs.
5962 * config/arm/arm.c (generic_addr_mode_costs): New.
5963 (arm_slowmul_tune): Initialise addr_mode_costs field.
5964 (arm_fastmul_tune): Likewise.
5965 (arm_strongarm_tune): Likewise.
5966 (arm_xscale_tune): Likewise.
5967 (arm_9e_tune): Likewise.
5968 (arm_marvell_pj4_tune): Likewise.
5969 (arm_v6t2_tune): Likewise.
5970 (arm_cortex_tune): Likewise.
5971 (arm_cortex_a8_tune): Likewise.
5972 (arm_cortex_a7_tune): Likewise.
5973 (arm_cortex_a15_tune): Likewise.
5974 (arm_cortex_a35_tune): Likewise.
5975 (arm_cortex_a53_tune): Likewise.
5976 (arm_cortex_a57_tune): Likewise.
5977 (arm_exynosm1_tune): Likewise.
5978 (arm_xgene1_tune): Likewise.
5979 (arm_cortex_a5_tune): Likewise.
5980 (arm_cortex_a9_tune): Likewise.
5981 (arm_cortex_a12_tune): Likewise.
5982 (arm_cortex_a73_tune): Likewise.
5983 (arm_v7m_tune): Likewise.
5984 (arm_cortex_m7_tune): Likewise.
5985 (arm_v6m_tune): Likewise.
5986 (arm_fa726te_tune): Likewise.
5987 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
5989 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5991 * config/arm/arm.c (arm_mem_costs): New function.
5992 (arm_rtx_costs_internal): Use arm_mem_costs.
5994 2017-11-23 Mark Wielaard <mark@klomp.org>
5996 * dwarf2out.c (init_sections_and_labels): Use generation to create
5997 unique ranges_section_label and ranges_base_label. Return generation.
5998 (output_rnglists): Add generation argument. Use generation to create
5999 unique ranges labels.
6000 (dwarf2out_finish): Get generation from init_sections_and_labels
6001 and pass generation to output_rnglists.
6003 2017-11-23 Mike Stump <mikestump@comcast.net>
6004 Eric Botcazou <ebotcazou@adacore.com>
6006 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
6007 * cfgloop.h (struct loop): Add unroll field.
6008 * function.h (struct function): Add has_unroll bitfield.
6009 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
6010 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
6011 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
6012 (pass_rtl_unroll_loops::gate): Likewise.
6013 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
6014 for which loop->unroll==1.
6015 (decide_unroll_constant_iterations): Use note for consistency and
6016 take loop->unroll into account. Return early if loop->unroll is set.
6017 Fix thinko in existing test.
6018 (decide_unroll_runtime_iterations): Use note for consistency and
6019 take loop->unroll into account.
6020 (decide_unroll_stupid): Likewise.
6021 * lto-streamer-in.c (input_cfg): Read loop->unroll.
6022 * lto-streamer-out.c (output_cfg): Write loop->unroll.
6023 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
6025 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
6026 (print_loop): Print loop->unroll if set.
6027 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
6028 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
6029 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
6031 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
6032 loop->unroll is set and smaller than the trip count. Otherwise bypass
6033 entirely the heuristics if loop->unroll is set. Remove dead note.
6034 Fix off-by-one bug in other note.
6035 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
6036 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
6038 (tree_unroll_loops_completely): Make static.
6039 (pass_complete_unroll::execute): Use correct type for variable.
6040 (pass_complete_unrolli::execute): Fix formatting.
6041 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
6043 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
6045 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
6046 enable when TARGET_PREFER_AVX128 is set.
6048 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6050 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
6052 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
6053 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
6055 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6057 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
6058 frequencies. Fix estimation of aggregate parameters.
6060 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6062 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
6065 2017-11-23 Tom de Vries <tom@codesourcery.com>
6067 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
6069 2017-11-23 Tom de Vries <tom@codesourcery.com>
6071 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
6074 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
6077 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
6078 sibcall_value_pcrel_fdpic): Use local variable instead of
6080 (calli_tbr_rel): Add missing operand 2.
6081 (call_valuei_tbr_rel): Add missing operand 3.
6083 2017-11-23 Jakub Jelinek <jakub@redhat.com>
6086 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
6087 bitpos/bitsize covering the whole destination, use store_expr only if
6088 the complex mode is the same. Otherwise, use expand_normal and if
6089 it returns CONCAT, subreg each part separately instead of trying to
6090 subreg the whole result.
6092 2017-11-23 Richard Biener <rguenther@suse.de>
6094 PR tree-optimization/23094
6095 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
6096 come here from walking over backedges in the first iteration.
6097 (vn_reference_lookup_3): Skip clobbers that store the same value.
6099 2017-11-23 Richard Biener <rguenther@suse.de>
6101 PR tree-optimization/81403
6102 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
6103 a block we need a leader relative to.
6104 (phi_translate_1): For nary processing require a leader from
6105 get_representative_for given we run expression simplification
6106 using match-and-simplify. Remove previous fix.
6108 2017-11-22 Jeff Law <law@redhat.com>
6110 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
6111 Use new method allocate_value_range rather than accessing the
6112 vrp_value_range_pool data member directly.
6113 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
6114 to use extract_range_from_stmt method to avoid need for
6115 extract_range_from_assignment method.
6116 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
6117 method rather than setting values_propgated data member directly.
6118 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
6119 and values propagated data members and extract_range_from_assignment
6120 method. Reorder private data members to conform to standards.
6121 Add new methods set_lattice_propagation_complete and
6122 allocate_value_range.
6124 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
6126 PR rtl-optimization/83030
6127 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
6128 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
6129 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
6130 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
6131 CROSSING_JUMP_P flag is already set before setting it.
6132 * cfgrtl.c (fixup_partition_crossing): Likewise.
6133 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
6136 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6138 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
6139 Handle the case where both arguments are using gen_const_vec_series.
6141 2017-11-22 David Malcolm <dmalcolm@redhat.com>
6144 * pretty-print.c (pp_format): Move quoting implementation to
6145 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
6146 to pass address of "quote" local.
6147 (pp_begin_quote): New function.
6148 (pp_end_quote): New function.
6149 * pretty-print.h (printer_fn): Convert penultimate param from bool
6151 (pp_begin_quote): New decl.
6152 (pp_end_quote): New decl.
6153 * tree-diagnostic.c (default_tree_printer): Convert penultimate
6154 param from bool to bool *.
6155 * tree-diagnostic.h (default_tree_printer): Likewise.
6157 2017-11-22 Jeff Law <law@redhat.com>
6159 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
6160 Initialize vr_values.
6161 (evrp_range_analyzer::try_find_new_range): Call methods attached to
6162 vr_values via vr_values class instance rather than delegators.
6163 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
6164 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
6165 (evrp_range_analyzer::push_value_range): Likewise.
6166 (evrp_range_analyzer::pop_value_range): Likewise.
6167 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
6168 most delegators. Those remaining are exposed as public interfaces.
6169 Make vr_values a pointer and private.
6170 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
6172 (evrp_range_analyzer::get_vr_value): New method.
6173 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
6174 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
6175 (class evrp_dom_walker): Attach evrp_folder class, initialize
6176 it in the ctor. Remove temporary delegators.
6177 (evrp_dom_walker::before_dom_children): Call methods in attached
6178 evrp_range_analyzer class via class instance pointer. Use
6179 free value_range_constant_singleton to remove need for
6180 op_with_constant_singleton_value delegator method. Do not
6181 create a vrp_prop class instance for every call! Narrow
6182 scope of a couple variables.
6183 (evrp_dom_walker::cleanup): Call methods in attached
6184 evrp_range_analyzer class via class instance pointer.
6185 * vr-values.h (class vr_values): Privatize many methods and
6188 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6190 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
6191 * cfgexpand.c (expand_debug_expr): Remove handling for them.
6192 * expr.c (expand_expr_real_2): Likewise.
6193 * fold-const.c (const_unop): Likewise.
6194 * optabs-tree.c (optab_for_tree_code): Likewise.
6195 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
6196 * tree-inline.c (estimate_operator_cost): Likewise.
6197 * tree-pretty-print.c (dump_generic_node): Likewise.
6198 (op_code_prio): Likewise.
6199 (op_symbol_code): Likewise.
6200 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
6201 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
6202 * internal-fn.c (direct_internal_fn_optab): New function.
6203 (direct_internal_fn_array, direct_internal_fn_supported_p
6204 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
6205 * fold-const-call.c (fold_const_reduction): New function.
6206 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
6208 * tree-vect-loop.c: Include internal-fn.h.
6209 (reduction_code_for_scalar_code): Rename to...
6210 (reduction_fn_for_scalar_code): ...this and return an internal
6212 (vect_model_reduction_cost): Take an internal_fn rather than
6214 (vect_create_epilog_for_reduction): Likewise. Build calls rather
6216 (vectorizable_reduction): Use internal functions rather than tree
6217 codes for the reduction operation. Update calls to the functions
6219 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
6220 Use calls to internal functions rather than REDUC tree codes.
6221 * config/aarch64/aarch64-simd.md: Update comment accordingly.
6223 2017-11-22 Olivier Hainque <hainque@adacore.com>
6225 * config/vxworks.c (vxworks_override_options): Pick default
6226 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
6227 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
6228 DWARF_GNAT_ENCODINGS_DEFAULT.
6229 * config/vxworksae.h: Likewise.
6231 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
6233 PR tree-optimization/83104
6234 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
6237 2017-11-22 Marek Polacek <polacek@redhat.com>
6238 H.J. Lu <hongjiu.lu@intel.com>
6239 Jason Merrill <jason@redhat.com>
6244 * calls.c (initialize_argument_information): Call
6245 warn_parameter_passing_abi target hook.
6246 (store_one_arg): Use 0 for empty record size. Don't push 0 size
6247 argument onto stack.
6248 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
6249 * common.opt: Update -fabi-version description.
6250 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
6251 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
6253 (ix86_is_empty_record): New function.
6254 (ix86_warn_parameter_passing_abi): New function.
6255 (TARGET_EMPTY_RECORD_P): Redefine.
6256 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
6257 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
6258 * doc/tm.texi: Regenerated.
6259 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
6260 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
6261 * dwarf2out.c (get_ultimate_context): Move to tree.c.
6262 * explow.c (hard_function_value): Call arg_int_size_in_bytes
6263 instead of int_size_in_bytes.
6264 * expr.c (copy_blkmode_to_reg): Likewise.
6265 * function.c (aggregate_value_p): Return 0 for empty types.
6266 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
6268 (locate_and_pad_parm): Call arg size_in_bytes instead
6270 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
6271 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
6272 * target.def (empty_record_p, warn_parameter_passing_abi): New target
6274 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
6275 (std_gimplify_va_arg_expr): Skip empty records. Call
6276 arg_size_in_bytes instead size_in_bytes.
6277 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
6278 * tree-core.h (tree_type_common): Add empty_flag.
6279 (tree_decl_common): Update comments.
6280 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
6282 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6283 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
6285 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6286 * tree.c (default_is_empty_type): New function.
6287 (default_is_empty_record): New function.
6288 (arg_int_size_in_bytes): New function.
6289 (arg_size_in_bytes): New function.
6290 (get_ultimate_context): New function.
6291 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
6292 TRANSLATION_UNIT_WARN_EMPTY_P.
6293 (default_is_empty_record, arg_int_size_in_bytes,
6294 arg_size_in_bytes, get_ultimate_context): Declare.
6296 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6298 * config/arm/arm.c (cmse_clear_registers): New function.
6299 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
6300 code by call to cmse_clear_registers.
6301 (cmse_nonsecure_entry_clear_before_return): Likewise.
6303 2017-11-22 Tamar Christina <tamar.christina@arm.com>
6305 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
6306 (vdot_s32, vdotq_s32): New.
6307 (vdot_lane_u32, vdotq_lane_u32): New.
6308 (vdot_lane_s32, vdotq_lane_s32): New.
6311 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6314 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
6315 for unsigned values with fewer HWIs than the precision.
6316 (test_overflow): New function.
6317 (wide_int_cc_tests): Call it.
6319 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6321 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
6322 has the same mode class as Pmode.
6324 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6326 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
6327 padding_bits_to_clear_ptr.
6328 (cmse_nonsecure_entry_clear_before_return): Likewise.
6330 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6332 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
6333 auto_sbitap instead of integer bitfield to control register needing
6336 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6338 PR tree-optimization/83044
6339 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
6340 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
6341 that need the upper bound. Subtract offset from
6342 get_addr_base_and_unit_offset only if positive and subtract it
6343 before division by eltsize rather than after it.
6346 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
6347 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
6348 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
6352 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
6353 expand_binop with *mul_widen_optab, make sure at least one of the
6354 operands doesn't have VOIDmode.
6357 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
6359 PR rtl-optimization/82044
6360 PR tree-optimization/82042
6361 * dse.c (record_store): Check for overflow.
6362 (check_mem_read_rtx): Properly check for overflow if width == -1, call
6363 add_wild_read instead of clear_rhs_from_active_local_stores on
6364 overflow and log it into dump_file.
6366 2017-11-22 Richard Biener <rguenther@suse.de>
6368 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
6369 fake edges to exit when looking for a place to insert.
6370 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
6372 (insert_into_preds_of_block): Commit edge insertion immediately,
6373 assert that doesn't require new BBs.
6374 (fini_pre): Release expressions.
6375 (pass_pre::execute): Shuffle things around a bit, if the fn
6376 is too large do not compute AVAIL either as this is really the
6379 2017-11-22 Richard Biener <rguenther@suse.de>
6381 PR tree-optimization/83089
6382 * tree-if-conv.c (pass_if_conversion::execute): If anything
6383 changed reset SCEV and free the number of iteration estimates.
6385 2017-11-21 Martin Sebor <msebor@redhat.com>
6387 PR tree-optimization/82945
6388 * calls.h (warn_nonstring_bound): Remove unused function.
6390 2017-11-21 Martin Sebor <msebor@redhat.com>
6392 PR tree-optimization/82945
6393 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
6394 * calls.h (maybe_warn_nonstring_arg): Declare new function.
6395 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
6397 (initialize_argument_information): Call maybe_warn_nonstring_arg.
6398 * calls.h (get_attr_nonstring_decl): Declare new function.
6399 * doc/extend.texi (attribute nonstring): Update.
6400 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
6401 get_attr_nonstring_decl and handle it.
6402 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
6403 detection of nul-termination.
6404 (strlen_to_stridx): Change to a pointer.
6405 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
6406 (pass_strlen::execute): Same.
6408 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
6410 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
6411 for the new option -mprefer-vector-width=[none|128|256|512].
6412 * config/i386/i386.c (ix86_target_string): remove old style options
6413 -mprefer-avx256 and make -mprefer-avx128 as alias.
6414 (ix86_option_override_internal): Apply defaults for the
6415 -mprefer-vector-width=[128|256] option.
6416 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
6417 Implement macros to work with -mprefer-vector-width=.
6418 * config/i386/i386.opt: Implemented option
6419 -mprefer-vector-width=[none|128|256|512].
6420 * doc/invoke.texi: Documentation for
6421 -mprefer-vector-width=[none|128|256|512].
6423 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
6425 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
6427 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
6429 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
6431 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
6433 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
6435 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
6436 * expr.c (expand_expr_real_2): Likewise.
6437 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
6438 fold_binary_loc): Likewise.
6439 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
6440 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
6441 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
6442 MINUS_EXPR transformations.
6443 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
6444 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
6445 * tree-inline.c (estimate_operator_cost): Likewise.
6446 * tree-pretty-print.c (dump_generic_node, op_code_prio,
6447 op_symbol_code): Likewise.
6448 * tree-vect-stmts.c (vectorizable_operation): Likewise.
6449 * vr-values.c (extract_range_from_binary_expr): Likewise.
6450 * varasm.c (initializer_constant_valid_p_1): Likewise.
6451 * tree.def: New tree code POINTER_DIFF_EXPR.
6453 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
6455 * config/i386/i386.md (*bswap<mode>2_movbe): Add
6456 integer suffix to movbe mnemonic.
6457 (*bswaphi2_movbe): Ditto.
6458 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
6460 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6463 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
6464 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
6465 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
6467 2017-11-21 Martin Liska <mliska@suse.cz>
6469 * tree-inline.c (expand_call_inline): Remove not needed
6472 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
6473 Jakub Jelinek <jakub@redhat.com>
6476 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
6477 Remove static keyword from f variable.
6479 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6481 PR tree-optimization/83086
6482 * gimple-ssa-store-merging.c
6483 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
6484 rather than n.base_addr.
6486 2017-11-21 Martin Liska <mliska@suse.cz>
6488 PR rtl-optimization/82044
6489 PR tree-optimization/82042
6490 * dse.c (check_mem_read_rtx): Check for overflow.
6492 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6494 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
6497 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6499 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
6500 and its corresponding call arg location note.
6502 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6503 Andrew Burgess <andrew.burgess@embecosm.com>
6505 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
6507 (arc_return_slot_offset): Likewise.
6508 (arc_eh_return_address_location): New declaration.
6509 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
6510 (MUST_SAVE_REGISTER): Add exception handler case.
6511 (MUST_SAVE_RETURN_ADDR): Likewise.
6512 (arc_frame_pointer_required): Likewise.
6513 (arc_frame_pointer_needed): New function.
6514 (arc_compute_frame_size): Changed.
6515 (arc_expand_prologue): Likewise.
6516 (arc_expand_epilogue): Likewise.
6517 (arc_initial_elimination_offset): Likewise.
6518 (arc_return_slot_offset): Delete.
6519 (arc_eh_return_address_location): New function.
6520 (arc_builtin_setjmp_frame_value): Likewise.
6521 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
6522 (EH_RETURN_STACKADJ_RTX): Define.
6523 (EH_RETURN_HANDLER_RTX): Likewise.
6524 * config/arc/arc.md (eh_return): Delete.
6526 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6528 * print-tree.h (debug_vec_tree): Remove prototype.
6529 * gdbinit.in (pvt): Remove macro.
6531 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6533 PR tree-optimization/83047
6534 * gimple-ssa-store-merging.c
6535 (imm_store_chain_info::output_merged_store): If the loads with the
6536 same vuse are in different basic blocks, for load_gsi pick a load
6537 location that is dominated by the other loads.
6540 * config/i386/i386.c (ix86_memmodel_check): Start
6541 -Winvalid-memory-model diagnostics with lowercase letter.
6544 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
6545 set in_cold_section_p to the partition containing loc_list->first.
6546 When seeing loc_list->last_before_switch node, update secname and
6547 perform range_across_switch second partition handling only after that.
6550 * run-rtl-passes.c: Include debug.h.
6551 (run_rtl_passes): Call debug_hooks->assembly_start.
6552 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
6556 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
6557 OPTAB_DIRECT in calls to expand_simple_binop.
6559 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6562 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
6564 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6567 * spellcheck.h (best_match::blithely_get_best_candidate): New
6570 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6572 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
6573 with lower case letter.
6575 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
6577 * config/i386/i386.md (bswaphi2): New expander.
6578 (*bswaphi2_movbe): New insn pattern.
6579 (bswaphi -> rorhi pepehole2): New peephole pattern.
6581 2017-11-20 Jeff Law <law@redhat.com>
6583 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
6584 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
6585 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
6586 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
6587 gimple-ssa-evrp-analyze.h.
6589 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
6590 set BB_VISITED here.
6591 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
6593 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6595 * tree-predcom.c: Add general comment on Store-Store chains.
6596 (split_data_refs_to_components): Postpone clearing eliminate_store_p
6598 (get_chain_last_ref_at): Rename into...
6599 (get_chain_last_write_at): ...this.
6600 (get_chain_last_write_before_load): New function.
6601 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
6602 CT_STORE_STORE when write reference is added.
6603 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
6604 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
6605 (initialize_root_vars_store_elim_1): Ditto.
6606 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
6607 definition is created.
6608 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
6609 chain by replacing it with dominant stored value.
6611 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6613 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
6615 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
6617 * vr-values.c (extract_range_from_binary_expr): Use a full range
6620 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
6622 * config/arm/arm.md (R4_REGNUM): Define constant.
6623 (nonsecure_call_internal): Remove r4 clobber.
6624 (nonsecure_call_value_internal): Likewise.
6625 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
6626 clobber and resequence match_operands.
6627 (nonsecure_call_value_reg_thumb1_v5): Likewise.
6628 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
6629 (nonsecure_call_value_reg_thumb2): Likewise.
6631 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6633 PR tree-optimization/78821
6634 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
6635 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
6636 to the address of the base rather than the base itself.
6637 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
6638 (find_bswap_or_nop_finalize): New function.
6639 (find_bswap_or_nop): Use it.
6640 (bswap_replace): Return a tree rather than bool, change first
6641 argument from gimple * to gimple_stmt_iterator, allow inserting
6642 into an empty sequence, allow ins_stmt to be NULL - then emit
6643 all stmts into gsi. Fix up MEM_REF address gimplification.
6644 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
6645 (struct store_immediate_info): Add N and INS_STMT non-static
6647 (store_immediate_info::store_immediate_info): Initialize them
6648 from newly added ctor args.
6649 (merged_store_group::apply_stores): Formatting fixes. Sort by
6651 (stmts_may_clobber_ref_p): For stores call also
6652 refs_anti_dependent_p.
6653 (gather_bswap_load_refs): New function.
6654 (imm_store_chain_info::try_coalesce_bswap): New method.
6655 (imm_store_chain_info::coalesce_immediate_stores): Use it.
6656 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
6657 (imm_store_chain_info::output_merged_store): Fail if number of
6658 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
6659 and NOP_EXPR rhs_code.
6660 (pass_store_merging::process_store): Compute n and ins_stmt, if
6661 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
6662 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
6664 (pass_store_merging::execute): Calculate dominators.
6666 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
6667 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
6668 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
6669 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6670 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
6671 class pass_optimize_bswap, bswap_replace,
6672 pass_optimize_bswap::execute): Moved to ...
6673 * gimple-ssa-store-merging.c: ... this file.
6674 Include optabs-tree.h.
6675 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
6676 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6677 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
6678 anonymous namespace, remove static keywords.
6679 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
6680 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
6682 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
6685 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
6687 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
6689 * vec.h (debug_helper): New function.
6690 (DEFINE_DEBUG_VEC): New macro.
6691 * hash-set.h (debug_helper): New function.
6692 (DEFINE_DEBUG_HASH_SET): New macro.
6693 * cfg.c (debug_slim (edge)): New function.
6694 Call DEFINE_DEBUG_VEC for edges.
6695 Call DEFINE_DEBUG_HASH_SET for edges.
6696 * cfghooks.c (debug_slim (basic_block)): New function.
6697 Call DEFINE_DEBUG_VEC for basic blocks.
6698 Call DEFINE_DEBUG_HASH_SET for basic blocks.
6699 * print-tree.c (debug_slim): New function to handle trees.
6700 Call DEFINE_DEBUG_VEC for trees.
6701 Call DEFINE_DEBUG_HASH_SET for trees.
6702 (debug (vec<tree, va_gc>) &): Remove.
6703 (debug (<vec<tree, va_gc>) *): Remove.
6704 * print-rtl.c (debug_slim): New function to handle const_rtx.
6705 Call DEFINE_DEBUG_VEC for rtx_def.
6706 Call DEFINE_DEBUG_VEC for rtx_insn.
6707 Call DEFINE_DEBUG_HASH_SET for rtx_def.
6708 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
6709 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
6710 (debug (vec<rtx_insn *> *ptr): Remove.
6711 (debug_insn_vector): Remove.
6712 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
6714 2017-11-20 Tom de Vries <tom@codesourcery.com>
6716 PR rtl-optimization/82020
6717 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
6718 IF_THEN_ELSE condition.
6720 2017-11-19 Jeff Law <law@redhat.com>
6722 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
6723 of degenerates resulting from ignoring an edge.
6725 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6728 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
6730 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6733 * profile-count.c (profile_count::to_sreal_scale): Fix return value
6734 for uninitialied counts.
6736 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6739 * tree-inline.c (estimate_num_insns): Set to 1 at least.
6741 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6744 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
6746 2017-11-19 Tom de Vries <tom@codesourcery.com>
6748 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
6751 2017-11-19 Tom de Vries <tom@codesourcery.com>
6753 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
6756 2017-11-19 Tom de Vries <tom@codesourcery.com>
6758 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6761 2017-11-19 Tom de Vries <tom@codesourcery.com>
6763 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
6765 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
6767 2017-11-19 Tom de Vries <tom@codesourcery.com>
6769 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
6771 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
6772 semicolon after MCORE_EXPORT_NAME call.
6774 2017-11-19 Tom de Vries <tom@codesourcery.com>
6777 * vmsdbgout.c (vmsdbgout_early_finish): New function.
6778 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
6780 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
6782 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
6783 (cgraph_edge::clone): Cleanup updating of profile.
6784 * ipa-cp.c (update_profiling_info): Likewise.
6785 * ipa-inline-transform.c (inline_transform): Likewise.
6786 * ipa-inline.c (inline_small_functions): Add missing space to dump.
6787 * ipa-split.c (execute_split_functions): Do not split when function
6789 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
6790 * profile-count.c (profile_count::dump): Add global0.
6791 (profile_count::to_cgraph_frequency): Do not ICE when entry is
6793 (profile_count::to_sreal_scale): Likewise.
6794 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
6795 (profile_count::combine_with_ipa_count): New function.
6796 * profile-count.h (profile_guessed_global0adjusted): New.
6797 (profile_count::adjusted_zero): New.
6798 (profile_count::global0adjusted): New.
6799 (profile_count::combine_with_ipa_count): New.
6800 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
6801 correct profile of return block of split functions.
6802 (copy_cfg_body): Remove unused profile_count.
6803 (copy_body): Likewise.
6804 (expand_call_inline): Update.
6805 (tree_function_versioning): Update.
6807 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6809 * hash-set.h (hash_set::empty): New.
6810 * tree-ssa-threadbackward.h: Delete.
6811 * tree-ssa-threadbackward.c (class thread_jumps): New.
6812 Move max_threaded_paths into class.
6813 (fsm_find_thread_path): Remove arguments that are now in class.
6814 (profitable_jump_thread_path): Rename to...
6815 (thread_jumps::profitable_jump_thread_path): ...this.
6816 (convert_and_register_jump_thread_path): Rename to...
6817 (thread_jumps::convert_and_register_current_path): ...this.
6818 (check_subpath_and_update_thread_path): Rename to...
6819 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6820 (register_jump_thread_path_if_profitable): Rename to...
6821 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6822 (handle_phi): Rename to...
6823 (thread_jumps::handle_phi): ...this.
6824 (handle_assignment): Rename to...
6825 (thread_jumps::handle_assignment): ...this.
6826 (fsm_find_control_statement_thread_paths): Rename to...
6827 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6828 (find_jump_threads_backwards): Rename to...
6829 (thread_jumps::find_jump_threads_backwards): ...this.
6830 Initialize path local data.
6831 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6832 from within thread_jumps class.
6833 (pass_early_thread_jumps::execute): Same.
6835 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6837 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6839 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6841 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6842 to canonical location.
6844 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6847 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6849 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6851 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6853 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6854 rcrt1.o%s/grcrt1.o%s for -static-pie.
6856 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6858 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6859 ix86_shift_rotate_cost): Break out from ...
6860 (ix86_rtx_costs): ... here.
6861 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6864 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6866 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6867 when entry block was promoted unlikely.
6868 (estimate_bb_frequencies): Increase frequency scale.
6869 * profile-count.h (profile_count): Export precision info.
6871 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6873 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6874 disturbin profile of entry block.
6876 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6878 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6881 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6883 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6884 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6886 2017-11-17 Jeff Law <law@redhat.com>
6888 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6889 from evrp_dom_walker class. Various methods moved into new class.
6890 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6891 (evrp_range_analyzer::enter): New method.
6892 (evrp_range_analyzer::leave): New method.
6893 (evrp_dom_walker): Remove delegators no longer needed by this class.
6894 Replace vr_values data member with evrp_range_analyzer
6896 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6897 method extracted from evrp_dom_walker::before_dom_children.
6898 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6899 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6901 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6902 Add private copy constructor and move assignment operators.
6903 Privatize methods and class data where trivially possible.
6904 (evrp_dom_walker::cleanup): New function, extracted from
6905 execute_early_vrp. Simplify access to class data.
6907 * vr-values.h (get_output_for_vrp): Prototype.
6908 * vr-values.c (get_output_for_vrp): New function extracted from
6909 vrp_visit_assignment_or_call and extract_range_from_stmt.
6910 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
6912 2017-11-17 Luis Machado <luis.machado@linaro.org>
6914 * config/aarch64/aarch64.c
6915 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
6916 (qdf24xx_tunings) <autoprefetcher_model>: Set to
6917 tune_params::AUTOPREFETCHER_WEAK.
6919 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6922 * config/arm/arm.c (arm_valid_target_attribute_rec):
6923 Parse "arch=" and "+<ext>".
6924 (arm_valid_target_attribute_tree): Re-init global options.
6925 (arm_option_override): Make non-static.
6926 (arm_options_perform_arch_sanity_checks): Make errors fatal.
6927 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
6928 (__ARM_FEATURE_CRC32): Support undef.
6929 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
6930 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
6932 2017-11-17 David Malcolm <dmalcolm@redhat.com>
6934 * gdbinit.in (break-on-diagnostic): New command.
6936 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6938 * config/i386/i386.c (ix86_expand_epilogue): Change simple
6939 return to indirect jump for EH return if control-flow
6940 protection is enabled. Change explicit 'false' argument in
6941 pro_epilogue_adjust_stack with a value of
6943 * config/i386/i386.md (simple_return_indirect_internal):
6944 Remove SImode restriction to support 64-bit.
6946 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6948 * combine.c (added_notes_insn): New.
6949 (try_combine): Handle added_notes_insn like added_links_insn.
6950 Rewrite return value code.
6951 (distribute_notes): Set added_notes_insn to the earliest insn we added
6954 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6956 PR rtl-optimization/82621
6957 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
6958 dest of one of those SETs is unused.
6960 2017-11-17 Richard Biener <rguenther@suse.de>
6963 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
6964 * tree-pretty-print.c (dump_generic_node): Handle
6965 annot_expr_parallel_kind.
6966 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
6967 * gimplify.c (gimple_boolify): Likewise.
6969 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6971 * config.gcc (extra_headers): Add cet.h for x86 targets.
6972 * config/i386/cet.h: New file.
6973 * doc/install.texi: Add --enable-cet/--disable-cet.
6975 2017-11-17 Richard Biener <rguenther@suse.de>
6977 PR tree-optimization/83017
6978 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
6979 (gen_parallel_loop): Properly count iterations.
6980 (parallelize_loops): Handle loop->can_be_parallel independent
6981 of flag_loop_parallelize_all. Make static profitability test match
6983 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
6984 * invoke.texi (parloops-min-per-thread): Document.
6986 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
6988 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
6989 upstreaming review comments.
6991 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6993 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
6994 with fast unaligned access.
6995 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
6997 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
6999 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
7000 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
7002 2017-11-17 Richard Biener <rguenther@suse.de>
7004 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
7005 folding of references.
7007 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
7010 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
7011 of replacing call to strncmp with corresponding call to strcmp when
7014 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
7016 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
7017 option prefer-avx256 for skylake-avx512 configuration.
7018 * config/i386/i386.c (ix86_option_override_internal): Ditto.
7019 (get_builtin_code_for_version): Ditto.
7021 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
7022 Monk Chiang <sh.chiang04@gmail.com>
7024 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
7025 (FIXED_REGISTERS): Reserve more register numbers.
7026 (CALL_USED_REGISTERS): Likewise.
7027 (REG_ALLOC_ORDER): Likewise.
7028 (REG_CLASS_CONTENTS): Likewise.
7029 (REGISTER_NAMES): Likewise.
7031 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
7032 Kito Cheng <kito.cheng@gmail.com>
7034 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
7036 * config/nds32/iterators.md: Add vector mode iterators and attributes.
7038 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
7040 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
7041 parameter list for vec_splats.
7043 2017-11-16 Joseph Myers <joseph@codesourcery.com>
7045 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
7047 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
7049 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
7052 * doc/install.texi: Document incompatibility of Perl >=5.6.26
7053 with the required version of automake 1.11.6.
7055 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
7057 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
7059 (DU_C2_3_power9): Correct reservation combinations.
7060 (FP_DIV_power9, VEC_DIV_power9): New.
7061 (power9-alu): Split out rotate/shift...
7062 (power9-rot): ...to here, correct dispatch resource.
7063 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
7065 (power9-fp): Correct latency.
7066 (power9-sdiv): Add div/sqrt resource.
7067 (power9-ddiv): Correct latency, add div/sqrt resource.
7068 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
7069 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
7071 (power9-qpdiv, power9-qpmul): Adjust resource usage.
7073 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
7075 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
7076 switch statement mapping KF built-ins to TF built-ins if we don't
7077 have the proper ISA 3.0 assembler support.
7079 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7081 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
7082 (lower_emutls_function_body): Do not compute it.
7084 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7086 * ipa-split.c (split_bb_info): Turn time to sreal.
7087 (split_point): Likewise.
7088 (dump_split_point): Likewise.
7089 (fine_split_points): Likewise.
7090 (execute_split_functions): Only zero split_bbs; turn time to sreals.
7092 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7094 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
7097 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7099 * predict.c (combine_predictions_for_bb): Preserve zero predicted
7101 (expensive_function_p): Remove useless assert.
7102 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
7104 2017-11-16 Martin Sebor <msebor@redhat.com>
7106 PR tree-optimization/82588
7107 PR tree-optimization/82583
7108 * tree-vrp.c (check_array_ref): Handle flexible array members,
7109 string literals, and inner indices.
7110 (search_for_addr_array): Add detail to diagnostics.
7112 2017-11-16 Nathan Sidwell <nathan@acm.org>
7116 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
7117 (SET_DECL_ASSEMBLER_NAME): Forward to
7118 overwrite_decl_assembler_name.
7119 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
7120 (overwrite_decl_assembler_name): Declare.
7121 * tree.c (overwrite_decl_assembler_name): New.
7122 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
7123 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
7124 (LANG_HOOKS_INITIALIZER): Add it.
7125 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
7126 * langhooks.c (lhd_set_decl_assembler_name): Use
7127 SET_DECL_ASSEMBLER_NAME.
7128 (lhd_overwrite_decl_assembler_name): Default implementation.
7130 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7131 Jackson Woodruff <jackson.woodruff@arm.com>
7133 PR tree-optimization/71026
7134 * match.pd: Canonicalize constant multiplies in division.
7136 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7138 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
7139 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
7140 Remove OPT_fomit_frame_pointer entry.
7141 * common/config/arc/arc-common.c: Likewise.
7142 * common/config/arm/arm-common.c: Likewise.
7143 * common/config/avr/avr-common.c: Likewise.
7144 * common/config/c6x/c6x-common.c: Likewise.
7145 * common/config/cr16/cr16-common.c: Likewise.
7146 * common/config/cris/cris-common.c: Likewise.
7147 * common/config/epiphany/epiphany-common.c: Likewise.
7148 * common/config/fr30/fr30-common.c: Likewise.
7149 * common/config/frv/frv-common.c: Likewise.
7150 * common/config/ia64/ia64-common.c: Likewise.
7151 * common/config/iq2000/iq2000-common.c: Likewise.
7152 * common/config/lm32/lm32-common.c: Likewise.
7153 * common/config/m32r/m32r-common.c: Likewise.
7154 * common/config/mcore/mcore-common.c: Likewise.
7155 * common/config/microblaze/microblaze-common.c: Likewise.
7156 * common/config/mips/mips-common.c: Likewise.
7157 * common/config/mmix/mmix-common.c: Likewise.
7158 * common/config/mn10300/mn10300-common.c: Likewise.
7159 * common/config/nios2/nios2-common.c: Likewise.
7160 * common/config/pa/pa-common.c: Likewise.
7161 * common/config/pdp11/pdp11-common.c: Likewise.
7162 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
7163 * common/config/riscv/riscv-common.c: Likewise.
7164 * common/config/rs6000/rs6000-common.c: Likewise.
7165 * common/config/rx/rx-common.c: Likewise.
7166 * common/config/s390/s390-common.c: Likewise.
7167 * common/config/sh/sh-common.c: Likewise.
7168 * common/config/sparc/sparc-common.c: Likewise.
7169 * common/config/tilegx/tilegx-common.c: Likewise.
7170 * common/config/tilepro/tilepro-common.c: Likewise.
7171 * common/config/v850/v850-common.c: Likewise.
7172 * common/config/visium/visium-common.c: Likewise.
7173 * common/config/xstormy16/xstormy16-common.c: Likewise.
7174 * common/config/xtensa/xtensa-common.c: Likewise.
7175 * invoke.texi (-fomit-frame-pointer): Update documentation.
7177 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7179 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
7181 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7183 * cfg.c (scale_bbs_frequencies_int,
7184 cale_bbs_frequencies_gcov_type): Remove.
7185 * cfg.h (scale_bbs_frequencies_int,
7186 cale_bbs_frequencies_gcov_type): Remove.
7188 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7190 * tree-ssa-loop-manip.c
7191 (scale_dominated_blocks_in_loop): Update to profile counts.
7192 (tree_transform_and_unroll_loop): Likewise.
7194 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7196 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
7197 scale_bbs_frequencies_int.
7199 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7201 * final.c (compute_alignments): Use counts rather than frequencies.
7203 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7205 * cfgloopanal.c: Include sreal.h
7206 (average_num_loop_insns): Use counts and sreal for accounting.
7208 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7210 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
7213 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7214 Kito Cheng <kito.cheng@gmail.com>
7216 * config/nds32/constraints.md: Provide more constraints.
7217 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
7218 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
7219 support constraints usage.
7221 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7223 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
7225 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7226 Kito Cheng <kito.cheng@gmail.com>
7228 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
7229 * config/nds32/nds32.opt: Refine the layout.
7230 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
7231 TARGET_EXT_STRING): Support new options.
7232 * config/nds32/nds32.h: Likewise.
7233 * config/nds32/nds32.md: Likewise.
7234 * config/nds32/nds32-predicates.c: Likewise.
7235 * config/nds32/constraints.md: Likewise.
7236 * common/config/nds32/nds32-common.c: Likewise.
7238 2017-11-16 Julia Koval <julia.koval@intel.com>
7241 * config/i386/gfniintrin.h: Add sse check.
7242 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
7244 2017-11-16 Julia Koval <julia.koval@intel.com>
7246 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
7247 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
7248 (ix86_handle_option): Handle -mavx512vbmi2.
7249 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
7250 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
7251 * config/i386/i386-c.c (__AVX512VBMI2__): New.
7252 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
7253 (ix86_valid_target_attribute_inner_p): Ditto.
7254 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
7255 * config/i386/i386.opt (mavx512vbmi2): New option.
7256 * doc/invoke.texi: Add new option.
7258 2017-11-16 Julia Koval <julia.koval@intel.com>
7260 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
7261 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
7262 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
7263 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
7264 _mm512_gf2p8mul_epi8): New intrinsics.
7265 * config/i386/i386-builtin-types.def
7266 (V64QI_FTYPE_V64QI_V64QI): New type.
7267 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
7268 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
7269 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
7270 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
7271 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
7272 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
7274 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
7276 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
7277 explicitly as a stream of bytes.
7279 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7281 * config/rs6000/altivec.h (vec_xst_be): New #define.
7282 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
7283 and externalize from *altivec_vperm_<mode>_internal.
7284 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
7286 (XL_BE_V8HI): Likewise.
7287 (XL_BE_V4SI): Likewise.
7288 (XL_BE_V4SI): Likewise.
7289 (XL_BE_V2DI): Likewise.
7290 (XL_BE_V4SF): Likewise.
7291 (XL_BE_V2DF): Likewise.
7292 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
7293 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
7294 all array entries with these keys: VSX_BUILTIN_VEC_XL,
7295 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
7296 VSX_BUILTIN_VEC_XST_BE.
7297 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
7298 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
7300 (altivec_init_builtins): Replace conditional calls to def_builtin
7301 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
7302 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
7303 with unconditional calls. Remove calls to def_builtin for
7304 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
7305 __builtin_vec_xst_be.
7306 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
7307 to define_expand, and add alternate RTL generation for P8.
7308 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
7309 vsx_ld_elemrev_v8hi.
7310 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
7311 add alternate RTL generation for P8.
7312 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
7313 vsx_ld_elemrev_v16qi.
7314 (vsx_st_elemrev_v8hi): Convert define_insn
7315 to define_expand, and add alternate RTL generation for P8.
7316 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
7317 vsx_st_elemrev_v8hi.
7318 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
7319 add alternate RTL generation for P8.
7320 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
7321 vsx_st_elemrev_v16qi.
7323 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
7326 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
7327 TARGET_AVX512ER check.
7328 (ix86_option_override_internal): Set MASK_VZEROUPPER if
7329 neither -mzeroupper nor -mno-zeroupper is used and
7330 TARGET_EMIT_VZEROUPPER is set.
7331 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
7332 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
7334 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
7336 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7337 folding of vector compares.
7338 (fold_build_vec_cmp): New helper function.
7339 (fold_compare_helper): New helper function.
7340 (builtin_function_type): Add compare builtins to the list of functions
7341 having unsigned arguments. Cosmetic updates to comment indentation.
7342 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
7343 the not+eq combination.
7345 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7347 PR tree-optimization/82726
7348 PR tree-optimization/70754
7349 * tree-predcom.c (order_drefs_by_pos): New function.
7350 (combine_chains): Move code setting has_max_use_after to...
7351 (try_combine_chains): ...here. New parameter. Sort combined chains
7352 according to position information.
7353 (tree_predictive_commoning_loop): Update call to above function.
7354 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
7356 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7358 PR tree-optimization/82726
7360 2017-01-23 Bin Cheng <bin.cheng@arm.com>
7362 PR tree-optimization/70754
7363 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
7364 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
7365 combined stmt before it if not NULL.
7366 (combine_chains): Process refs reversely and compute dominance point
7370 2017-02-23 Bin Cheng <bin.cheng@arm.com>
7372 PR tree-optimization/79663
7373 * tree-predcom.c (combine_chains): Process refs in reverse order
7374 only for ZERO length chains, and add explaining comment.
7376 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7378 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
7379 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
7380 * doc/invoke.texi (ARM Options): Add armv8.3-a.
7382 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7384 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
7386 2017-11-15 Martin Liska <mliska@suse.cz>
7388 * tree-cfg.c (pass_warn_function_return::execute):
7389 Compare warn_return_type for greater than zero.
7391 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
7395 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
7396 to return true on Xeon and not on Xeon Phi.
7397 (ix86_check_avx256_register): Changed to ...
7398 (ix86_check_avx_upper_register): ... this. Add extra check for
7399 VALID_AVX512F_REG_OR_XI_MODE.
7400 (ix86_avx_u128_mode_needed): Changed
7401 ix86_check_avx256_register to ix86_check_avx_upper_register.
7402 (ix86_check_avx256_stores): Changed to ...
7403 (ix86_check_avx_upper_stores): ... this. Changed
7404 ix86_check_avx256_register to ix86_check_avx_upper_register.
7405 (ix86_avx_u128_mode_after): Changed
7406 avx_reg256_found to avx_upper_reg_found. Changed
7407 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
7408 (ix86_avx_u128_mode_entry): Changed
7409 ix86_check_avx256_register to ix86_check_avx_upper_register.
7410 (ix86_avx_u128_mode_exit): Ditto.
7411 * config/i386/i386.h: (host_detect_local_cpu): New define.
7413 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7415 * config/arm/xgene1.md (xgene1): Split into automatons
7416 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
7417 (xgene1_f_load): Adjust reservations and/or types.
7418 (xgene1_f_store): Likewise.
7419 (xgene1_load_pair): Likewise.
7420 (xgene1_store_pair): Likewise.
7421 (xgene1_fp_load1): Likewise.
7422 (xgene1_load1): Likewise.
7423 (xgene1_store1): Likewise.
7424 (xgene1_move): Likewise.
7425 (xgene1_alu): Likewise.
7426 (xgene1_simd): Likewise.
7427 (xgene1_bfm): Likewise.
7428 (xgene1_neon_load1): Likewise.
7429 (xgene1_neon_store1): Likewise.
7430 (xgene1_neon_logic): Likewise.
7431 (xgene1_neon_st1): Likewise.
7432 (xgene1_neon_ld1r): Likewise.
7433 (xgene1_alu_cond): Added.
7434 (xgene1_shift_reg): Likwise.
7435 (xgene1_bfx): Likewise.
7436 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
7438 2017-11-15 Jakub Jelinek <jakub@redhat.com>
7441 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
7443 (can_widen_mult_without_libcall): New function.
7444 (expand_mul_overflow): If only checking unsigned mul overflow,
7445 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
7446 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
7447 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
7448 (expand_DIVMOD): Formatting fix.
7449 * expmed.h (expand_mult): Add NO_LIBCALL argument.
7450 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
7451 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
7453 PR tree-optimization/82977
7454 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
7455 constructed temporary to strlen_to_stridx.put.
7457 2017-11-15 Martin Liska <mliska@suse.cz>
7459 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
7460 * configure: Regenerate.
7462 2017-11-15 Martin Liska <mliska@suse.cz>
7465 * config/sh/sh-mem.cc: Use proper probability for
7468 2017-11-14 Jeff Law <law@redhat.com>
7470 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
7471 the red zone for stack_clash_protection_final_dynamic_probe targets
7472 when the total dynamic stack size is zero bytes.
7474 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
7475 blocks is post order.
7477 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
7479 * dumpfile.h (TDF_COMPARE_DEBUG): New.
7480 * final.c (rest_of_clean_state): Set it for the
7481 -fcompare-debug dump.
7482 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
7483 class when TDF_COMPARE_DEBUG is set.
7485 * dwarf2out.c (gen_producer_string): Discard
7488 2017-11-15 Joseph Myers <joseph@codesourcery.com>
7491 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
7492 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
7493 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
7494 (__TGMATH_REAL_2_3): Remove macros.
7495 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
7496 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
7498 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
7500 (remquo): Define using __TGMATH_REAL_3.
7502 2017-11-14 Jeff Law <law@redhat.com>
7504 * vr-values.c: New file with contents extracted from tree-vrp.c.
7505 * Makefile.in (OBJS): Add vr-values.o
7506 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
7507 (set_value_range, set_and_canonicalize_value_range): Likewise.
7508 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
7509 (value_range_constant_singleton, symbolic_range_p): Likewise.
7510 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
7511 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
7512 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
7513 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
7514 (find_case_label_range, find_case_label_index): Likewise.
7515 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
7516 (range_int_cst_singleton_p, value_inside_range): Likewise.
7517 (get_single_symbol): Likewise.
7518 (switch_update): Move structure definition here.
7519 (to_remove_edges, to_update_switch_stmts): Provide externs.
7520 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
7521 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
7522 (vrp_val_is_min, set_value_range): Likewise.
7523 (set_and_canonicalize_value_range, copy_value_range): Likewise.
7524 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
7525 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
7526 (range_is_nonnull, range_int_cst_p): Likewwise.
7527 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
7528 (get_single_symbol, operand_less_p): Likewise
7529 (compare_values_warnv, compare_values): Likewise.
7530 (value_inside_range, value_range_constant_singleton): Likewise.
7531 (zero_nonzero_bitgs_from_vr): Likewise.
7532 (extract_range_from_binary_expr_1): Likewise.
7533 (overflow_comparison_p): Likewise.
7534 (to_remove_edges, to_update_switch_stmts): Likewise.
7535 (find_case_label-index, find_case_label_range): Likewise.
7536 (switch_update, set_value_range_to_nonnegative): Remove.
7537 (set_value_range_to_truthvalue): Likewise.
7538 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
7539 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
7540 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
7541 (find_case_label_ranges, test_for_singularity): Likewise.
7542 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
7543 (x_vr_values): Move to its remaining use site.
7545 2017-11-10 Jeff Law <law@redhat.com>
7547 * vr-values.h (VR_INITIALIZER): Move #define here.
7548 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
7549 * Makefile.in (OBJS): Add tree-evrp.o
7550 * tree-vrp.h (assert_info): Move structure definition here.
7551 (set_value_range_to_varying): Prototype.
7552 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7553 (infer_value_range, register_edge_assert_for): Likewise.
7554 (stmt_interesting_for_vrp): Likewise.
7555 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
7556 (set_value_range_to_varying): No longer static.
7557 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7558 (infer_value_range, register_edge_assert_for): Likewise.
7560 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7562 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
7563 generate the XXBRD instruction.
7565 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
7566 (rs6000_builtin_type_compatible): Treat _Float128 and long double
7567 as being compatible if -mabi=ieeelongdouble.
7568 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
7569 to setup float128 built-ins with hardware support.
7570 (BU_FLOAT128_HW_2): Likewise.
7571 (BU_FLOAT128_HW_3): Likewise.
7572 (BU_FLOAT128_HW_VSX_1): Likewise.
7573 (BU_FLOAT128_HW_VSX_2): Likewise.
7574 (scalar_extract_expq): Change float128 built-in functions to
7575 accommodate having both KFmode and TFmode functions. Use the
7576 KFmode variant as the default.
7577 (scalar_extract_sigq): Likewise.
7578 (scalar_test_neg_qp): Likewise.
7579 (scalar_insert_exp_q): Likewise.
7580 (scalar_insert_exp_qp): Likewise.
7581 (scalar_test_data_class_qp): Likewise.
7582 (sqrtf128_round_to_odd): Delete processing the round to odd
7583 built-in functions as special built-in functions, and define them
7584 as float128 built-ins. Use the KFmode variant as the default.
7585 (truncf128_round_to_odd): Likewise.
7586 (addf128_round_to_odd): Likewise.
7587 (subf128_round_to_odd): Likewise.
7588 (mulf128_round_to_odd): Likewise.
7589 (divf128_round_to_odd): Likewise.
7590 (fmaf128_round_to_odd): Likewise.
7591 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7592 support for KFmode and TFmode xststdcqp calls.
7593 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
7594 point, switch the built-in handlers for the get/set float128
7595 exponent, get float128 mantissa, float128 test built-ins, and the
7596 float128 round to odd built-in functions. Eliminate creating the
7597 float128 round to odd built-in functions as special built-ins.
7598 (rs6000_init_builtins): Eliminate special creation of the float128
7599 round to odd built-in functions.
7600 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
7601 function insns to support both TFmode and KFmode variants.
7602 (xsxsigqp_<mode>): Likewise.
7603 (xsiexpqpf_<mode>): Likewise.
7604 (xsiexpqp_<mode>): Likewise.
7605 (xststdcqp_<mode>): Likewise.
7606 (xststdcnegqp_<mode>): Likewise.
7607 (xststdcqp_<mode>): Likewise.
7609 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7611 * tree-ssa-threadupdate.c (compute_path_counts): Remove
7612 unused path_in_freq_ptr parameter.
7613 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
7615 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7617 * ipa-inline.c (edge_badness): Dump sreal frequency.
7618 (compute_inlined_call_time): Match natural implementaiton ...
7619 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
7620 forgotten division by CGRAPH_FREQ_BASE.
7622 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7624 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
7625 Solaris 11. Update comment.
7626 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
7628 * config/sol2.h (STARTFILE_SPEC): Likewise.
7629 * configure: Regenerate.
7631 2017-11-14 Carl Love <cel@us.ibm.com>
7633 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
7634 le_ and be_ prefixes to swap* variables. Remove
7635 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
7638 2017-11-14 Jason Merrill <jason@redhat.com>
7640 Support GTY((cache)) on hash_map.
7641 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
7642 (ggc_cache_remove): Override it instead of ggc_mx.
7643 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
7644 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
7645 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
7646 (simple_cache_map_traits): Override maybe_mx.
7647 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
7648 (hash_map): Friend gt_cleare_cache.
7649 (gt_cleare_cache): New.
7650 * tree.h (tree_cache_traits): New hash_map traits class.
7651 (tree_cache_map): New typedef.
7653 2017-11-14 Richard Biener <rguenther@suse.de>
7655 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
7656 paramter and handling.
7657 (cleanup_control_flow_bb): Likewise.
7658 (cleanup_control_flow_pre): New helper performing a DFS walk
7659 to call cleanup_control_flow_bb in PRE order.
7660 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
7661 via cleanup_control_flow_pre.
7663 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
7665 * config/aarch64/aarch64-simd.md
7666 (aarch64_simd_bsl<mode>_internal): Remove DImode.
7667 (*aarch64_simd_bsl<mode>_alt): Likewise.
7668 (aarch64_simd_bsldi_internal): New.
7669 (aarch64_simd_bsldi_alt): Likewise.
7671 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7673 * tracer.c (better_p): Do not compare frequencies.
7674 * reg-stack.c (better_edge): Likewise.
7675 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
7678 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7680 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
7681 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
7682 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
7683 * lto-streamer-in.c (input_function): Use update_max_bb_count.
7684 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
7685 * predict.c (maybe_hot_frequency_p): Inline to ...
7686 (maybe_hot_count_p): ... here; rewrite to counts.
7687 (counts_to_freqs): Rename to ...
7688 (update_max_bb_count): ... this one.
7689 (expensive_function_p): Use counts.
7690 (estimate_bb_frequencies): Update.
7691 (rebuild_frequencies): Update.
7692 * predict.h (counts_to_freqs): Rename to ...
7693 (update_max_bb_count): ... this one.
7694 * profile.c (compute_branch_probabilities): Add debug info
7695 * tree-inline.c (expand_call_inline): Update debug info.
7696 (optimize_inline_calls): Use update_max_bb_count..
7697 (tree_function_versioning): Use update_max_bb_count..
7698 * value-prof.c (gimple_value_profile_transformations):
7699 Do not use update_max_bb_count.
7701 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7703 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
7704 always use frequencies.
7706 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7708 * bb-reorder.c: Remove frequencies from comments.
7709 (better_edge_p): Use profile counts.
7710 (find_traces): Dump profile counts.
7711 (rotate_loop): Use profile counts.
7712 (find_traces_1_round): Likewise.
7713 (connect_better_edge_p): Use counts instead of probabilities for
7715 (copy_bb_p): Drop early check for non-0 frequency.
7716 (sanitize_hot_paths): Update comments.
7718 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7720 * ipa-split.c (struct split_point): Add count.
7721 (consider_split): Do not compute incoming frequency; compute incoming
7722 count and store it to split_point.
7723 (split_function): Set count of the call to split part correctly.
7725 2017-11-13 Carl Love <cel@us.ibm.com>
7727 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
7729 2017-11-13 Tom Tromey <tom@tromey.com>
7731 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
7733 2017-11-13 Carl Love <cel@us.ibm.com>
7735 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
7736 Add support for builtins:
7737 unsigned int vec_first_{,miss}_match_{,or_eos}index,
7738 vector {un,}signed {char,int,short},
7739 vector {un,}signed {char,int,short}) arguments.
7740 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
7741 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
7742 Add BU_P9V_AV_2 expansions for the builtins.
7743 * config/rs6000/altivec.h (vec_first_match_index,
7744 vec_first_mismatch_index, vec_first_match_or_eos_index,
7745 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
7746 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
7747 new extern declaration.
7748 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
7749 * config/rs6000/vsx.md (first_match_index_<mode>,
7750 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
7751 first_mismatch_or_eos_index_<mode>): Add define expand.
7752 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
7753 * doc/extend.texi: Update the built-in documenation file for the new
7756 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
7758 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
7759 into the min/max operations for _Float<N> and _Float<N>X types.
7761 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7764 * dwarf2out.c (do_eh_frame): New static variable.
7765 (dwarf2out_begin_prologue): Set it.
7766 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
7768 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7770 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
7772 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
7774 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7777 * debug.h (dwarf2out_do_eh_frame): Declare.
7778 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
7779 (dwarf2out_do_frame): Use it.
7780 (dwarf2out_do_cfi_asm): Likewise.
7781 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
7782 (dwarf2out_assembly_start): Likewise.
7783 (dwarf2out_begin_prologue): Fix comment.
7784 * toplev.c (compile_file): Always call dwarf2out_frame_finish
7785 if the target needs either debug or unwind DWARF2 info.
7786 * lto-opts.c (lto_write_options): Do not save -fexceptions,
7787 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
7788 -ftrapping-math, -ftrapv and -fwrapv.
7790 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7792 * cgraph.c (cgraph_edge::sreal_frequency): New function.
7793 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
7794 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
7795 (estimate_edge_size_and_time): Likewise.
7796 (ipa_merge_fn_summary_after_inlining): Likewise.
7797 * ipa-inline.c (cgraph_freq_base_rec): Remove.
7798 (compute_uninlined_call_time): Use sreal_frequency.
7799 (compute_inlined_call_time): Likewise.
7800 (ipa_inline): Do not initialize cgraph_freq_base_rec.
7801 * profile-count.c: Include sreal.h.
7802 (profile_count::to_sreal_scale): New.
7803 * profile-count.h: Forward declare sreal.
7804 (profile_count::to_sreal_scale): Declare.
7806 2017-11-13 Nathan Sidwell <nathan@acm.org>
7808 * diagnostic.c (maybe_line_and_column): New.
7809 (diagnostic_get_location_text): Use it.
7810 (diagnostic_report_current_module): Likewise.
7811 (test_diagnostic_get_location_text): Add tests.
7813 2017-11-13 Luis Machado <luis.machado@linaro.org>
7815 * doc/md.texi (Specifying processor pipeline description): Fix
7816 incorrect latency for the div instruction example.
7818 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7820 PR tree-optimization/78821
7821 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7822 that bit_not_p is the same.
7823 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7824 (split_group): Count precisely bit_not_p bits in each statement.
7825 (invert_op): New function.
7826 (imm_store_chain_info::output_merged_store): Use invert_op to
7827 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7828 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7830 2017-11-13 Martin Liska <mliska@suse.cz>
7832 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7833 (struct source_info): Likewise.
7834 (add_branch_counts): Likewise.
7835 (add_line_counts): Likewise.
7836 (function_summary): Likewise.
7837 (output_intermediate_line): Likewise.
7838 (generate_results): Likewise.
7840 2017-11-13 Martin Liska <mliska@suse.cz>
7842 * gcov.c (struct block_info): Remove typedef for block_t.
7843 (struct line_info): Likewise.
7844 (line_info::has_block): Likewise.
7845 (EXIT_BLOCK): Likewise.
7846 (unblock): Likewise.
7847 (circuit): Likewise.
7848 (get_cycles_count): Likewise.
7849 (process_file): Likewise.
7850 (read_graph_file): Likewise.
7851 (solve_flow_graph): Likewise.
7852 (find_exception_blocks): Likewise.
7853 (add_line_counts): Likewise.
7854 (accumulate_line_info): Likewise.
7855 (output_line_details): Likewise.
7857 2017-11-13 Martin Liska <mliska@suse.cz>
7859 * gcov.c (struct arc_info): Remove typedef for arc_t.
7860 (struct line_info): Likewise.
7861 (add_branch_counts): Likewise.
7862 (output_branch_count): Likewise.
7863 (function_info::~function_info): Likewise.
7864 (circuit): Likewise.
7865 (output_intermediate_line): Likewise.
7866 (read_graph_file): Likewise.
7867 (solve_flow_graph): Likewise.
7868 (find_exception_blocks): Likewise.
7869 (add_line_counts): Likewise.
7870 (accumulate_line_info): Likewise.
7871 (output_line_details): Likewise.
7872 (output_function_details): Likewise.
7874 2017-11-13 Martin Liska <mliska@suse.cz>
7876 * gcov.c (struct function_info): Remove typedef for function_t.
7877 (struct source_info): Likewise.
7878 (source_info::get_functions_at_location): Likewise.
7879 (solve_flow_graph): Likewise.
7880 (find_exception_blocks): Likewise.
7881 (add_line_counts): Likewise.
7882 (output_intermediate_file): Likewise.
7883 (process_file): Likewise.
7884 (generate_results): Likewise.
7885 (release_structures): Likewise.
7886 (read_graph_file): Likewise.
7887 (read_count_file): Likewise.
7888 (accumulate_line_counts): Likewise.
7889 (output_lines): Likewise.
7891 2017-11-13 Martin Liska <mliska@suse.cz>
7893 * gcov.c (function_info::function_info): Remove num_counts
7894 and add vector<gcov_type>.
7895 (function_info::~function_info): Use the vector.
7896 (process_file): Likewise.
7897 (read_graph_file): Likewise.
7898 (read_count_file): Likewise.
7899 (solve_flow_graph): Likewise.
7901 2017-11-13 Martin Liska <mliska@suse.cz>
7903 * gcov.c (function_info::is_artificial): New function.
7904 (process_file): Erase all artificial early.
7905 (generate_results): Skip as all artificial are already
7908 2017-11-13 Martin Liska <mliska@suse.cz>
7910 * gcov.c (read_graph_file): Store to global vector of functions.
7911 (read_count_file): Iterate the vector.
7912 (process_file): Likewise.
7913 (generate_results): Likewise.
7914 (release_structures): Likewise.
7916 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7918 PR tree-optimization/82954
7919 * gimple-ssa-store-merging.c
7920 (imm_store_chain_info::coalesce_immediate_stores): If
7921 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
7923 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
7925 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7926 Upddate call to ENDIAN_LANE_N.
7927 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
7928 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
7929 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
7930 and use aarch64_endian_lane_rtx.
7931 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
7933 2017-11-12 Tom de Vries <tom@codesourcery.com>
7935 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
7937 2017-11-12 Tom de Vries <tom@codesourcery.com>
7939 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
7941 2017-11-12 Tom de Vries <tom@codesourcery.com>
7943 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
7945 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7946 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7947 * defaults.h (ASM_OUTPUT_LABELREF): Same.
7949 2017-11-11 Martin Sebor <msebor@redhat.com>
7952 * doc/extend.texi (attribute nonstring): Remove spurious argument.
7955 * prefic.c (translate_name): Replace strncpy with memcpy to
7956 avoid -Wstringop-truncation.
7958 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7960 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
7962 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7964 * predict.c (maybe_hot_frequency_p): Do not use cfun.
7966 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7968 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
7971 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7973 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
7974 enable generating XXBRH if the value is in a vector register.
7975 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
7976 value is in a vector register.
7977 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
7978 register to register bswap64's instead of doing the GPR sequence
7979 used on previous machines.
7980 (bswapdi2_xxbrd): New insn.
7981 (bswapdi2_reg): Disallow on ISA 3.0.
7982 (register to register bswap64 splitter): Do not split the insn on
7983 ISA 3.0 systems that use XXBRD.
7985 2017-11-10 Martin Sebor <msebor@redhat.com>
7988 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
7990 (find_subframework_file): Same.
7992 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7994 * auto-profile.c (afdo_indirect_call): Drop frequency.
7995 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
7996 (cgraph_node::create_edge): Drop frequency argument.
7997 (cgraph_node::create_indirect_edge): Drop frequency argument.
7998 (cgraph_edge::make_speculative): Drop frequency arguments.
7999 (cgraph_edge::resolve_speculation): Do not update frequencies
8000 (cgraph_edge::dump_edge_flags): Do not dump frequency.
8001 (cgraph_node::dump): Check consistency in IPA mode.
8002 (cgraph_edge::maybe_hot_p): Use IPA counter.
8003 (cgraph_edge::verify_count_and_frequency): Rename to ...
8004 (cgraph_edge::verify_count): ... this one; drop frequency checking.
8005 (cgraph_node::verify_node): Update.
8006 * cgraph.h (struct cgraph_edge): Drop frequency.
8007 (cgraph_edge::frequency): New function.
8008 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
8010 (cgraph_edge::rebuild_edges): Likewise.
8011 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
8012 (duplicate_thunk_for_node): Do not pass frequency.
8013 (cgraph_node::create_clone): Scale only counts.
8014 (cgraph_node::create_virtual_clone): Do not pass frequency.
8015 (cgraph_node::create_edge_including_clones): Do not pass frequency.
8016 (cgraph_node::create_version_clone): Do not pass frequency.
8017 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
8018 (cgraph_node::expand_thunk): Do not pass frequency.
8019 (cgraph_node::create_wrapper): Do not pass frequency.
8020 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
8022 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
8023 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
8024 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
8025 (gather_caller_stats): Use frequency function.
8026 (ipcp_cloning_candidate_p): Use frequency function.
8027 (ipcp_propagate_stage): Use frequency function.
8028 (get_info_about_necessary_edges): Use frequency function.
8029 (update_profiling_info): Update only IPA profile.
8030 (update_specialized_profile): Use frequency functoin.
8031 (perhaps_add_new_callers): Update only IPA profile.
8032 * ipa-devirt.c (ipa_devirt): Use IPA profile.
8033 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
8034 (dump_ipa_call_summary): Use frequency function.
8035 (estimate_edge_size_and_time): Use frequency function.
8036 (ipa_merge_fn_summary_after_inlining): Use frequency function.
8037 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
8038 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
8039 (update_noncloned_counts): ... ths one; scale counts only.
8040 (clone_inlined_nodes): Do not scale frequency.
8041 (inline_call): Do not pass frequency.
8042 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
8043 (compute_inlined_call_time): Use IPA profile.
8044 (want_inline_small_function_p): Use IPA profile.
8045 (want_inline_self_recursive_call_p): Use IPA profile.
8046 (edge_badness): Use IPA profile.
8047 (lookup_recursive_calls): Use IPA profile.
8048 (recursive_inlining): Do not pass frequency.
8049 (resolve_noninline_speculation): Do not update frequency.
8050 (inline_small_functions): Collect max of IPA profile.
8051 (dump_overall_stats): Dump IPA porfile.
8052 (dump_inline_stats): Dump IPA porfile.
8053 (ipa_inline): Collect IPA stats.
8054 * ipa-inline.h (clone_inlined_nodes): Update prototype.
8055 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
8056 (ipa_propagate_frequency): Use frequency function.
8057 (ipa_profile): Cleanup.
8058 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
8059 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
8060 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
8061 (input_node): Do not stream frequency.
8062 (input_edge): Do not stream frequency.
8063 (merge_profile_summaries): Scale only IPA profiles.
8064 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
8065 * predict.c (drop_profile): Do not recompute frequency.
8066 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
8067 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
8068 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
8069 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
8070 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
8071 * tree-inline.c (copy_bb): Do not scale frequency.
8072 (expand_call_inline): Do not scale frequency.
8073 (tree_function_versioning): Do not scale frequency.
8074 * ubsan.c (ubsan_create_edge): Do not pass frequency.
8076 2017-11-10 Julia Koval <julia.koval@intel.com>
8078 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
8079 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
8080 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
8081 (_mm256_maskz_gf2p8affine_epi64_epi8)
8082 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
8083 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
8084 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
8085 (__builtin_ia32_vgf2p8affineqb_v32qi)
8086 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
8087 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
8089 2017-11-10 Tamar Christina <tamar.christina@arm.com>
8093 (arm_option_override): Refactor.
8094 (arm_option_reconfigure_globals): New.
8095 (arm_options_perform_arch_sanity_checks): New.
8096 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
8098 (arm_options_perform_arch_sanity_checks): Likewise
8100 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
8102 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
8103 (power9-qpmul): New.
8104 * rs6000/rs6000.md ("type" attr): Add qmul.
8105 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
8106 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
8107 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
8109 2017-11-10 Martin Sebor <msebor@redhat.com>
8112 * builtins.c (compute_objsize): Handle arrays that
8113 compute_builtin_object_size likes to fail for. Make extern.
8114 * builtins.h (compute_objsize): Declare.
8115 (check_strncpy_sizes): New function.
8116 (expand_builtin_strncpy): Call check_strncpy_sizes.
8117 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
8118 -Wstringop-truncation.
8119 (gimple_fold_builtin_strncat): Same.
8120 * gimple.c (gimple_build_call_from_tree): Set call location.
8121 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
8122 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
8123 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
8124 (handle_builtin_strlen): Use strlen_to_stridx.
8125 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
8127 Use strlen_to_stridx.
8128 (pass_strlen::execute): Release strlen_to_stridx.
8129 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
8130 (-Wstringop-truncation): Document new option.
8132 2017-11-10 Martin Liska <mliska@suse.cz>
8134 PR gcov-profile/82702
8135 * gcov.c (main): Handle intermediate files in a different
8137 (get_gcov_intermediate_filename): New function.
8138 (output_gcov_file): Remove support of intermediate files.
8139 (generate_results): Allocate intermediate file.
8140 (release_structures): Clean-up properly fn_end.
8141 (output_intermediate_file): Start iterating with line 1.
8143 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8145 PR tree-optimization/82929
8146 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8147 ops_swapped_p non-static data member.
8148 (store_immediate_info::store_immediate_info): Clear it.
8149 (imm_store_chain_info::coalesce_immediate_stores): If swapping
8150 ops set ops_swapped_p.
8151 (count_multiple_uses): Handle ops_swapped_p.
8153 2017-11-10 Martin Liska <mliska@suse.cz>
8155 * coverage.c (coverage_init): Stream information about
8156 support of has_unexecuted_blocks.
8157 * doc/gcov.texi: Document that.
8158 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
8159 * gcov.c (read_graph_file): Likewise.
8160 (output_line_beginning): Fix a small issue with
8163 2017-11-10 Bin Cheng <bin.cheng@arm.com>
8165 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
8166 reference of trivial component.
8168 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8171 * gimple-ssa-store-merging.c
8172 (pass_store_merging::terminate_all_aliasing_chains): For
8173 gimple_store_p stmts also call refs_output_dependent_p.
8175 PR rtl-optimization/82913
8176 * compare-elim.c (try_merge_compare): Punt if def_insn is not
8179 2017-11-09 Jeff Law <law@redhat.com>
8181 * vr-values.h: New file with vr_values class.
8182 * tree-vrp.c: Include vr-values.h
8183 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
8184 data objects into the vr_values class.
8185 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
8186 (get_value_range): Make it a member function within vr_values class.
8187 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
8188 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
8189 (op_with_constant_singleton_value_range): Likewise.
8190 (extract_range_for_var_from_comparison_expr): Likewise.
8191 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
8192 (extract_range_from_binary_expr): Likewise.
8193 (extract_range_from_unary_expr): Likewise.
8194 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
8195 (check_for_binary_op_overflow, extract_range_basic): Likewise.
8196 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
8197 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
8198 (compare_name_with_value, compare_names): Likewise.
8199 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
8200 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
8201 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
8202 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
8203 (extract_range_from_phi_node): Likewise.
8204 (simplify_truth_ops_using_ranges): Likewise.
8205 (simplify_div_or_mod_using_ranges): Likewise.
8206 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
8207 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
8208 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
8209 (simplify_float_conversion_using_ranges): Likewise.
8210 (simplify_internal_call_using_ranges): Likewise.
8211 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
8212 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
8213 poitner via x_vr_values for calls into gimple folder.
8214 (vrp_initialize_lattice): Make this the vr_values ctor.
8215 (vrp_free_lattice): Make this the vr_values dtor.
8216 (set_vr_value): New function.
8217 (class vrp_prop): Add vr_values data member. Add various member
8218 functions as well as member functions that delegate to vr_values.
8219 (check_array_ref): Make a member function within vrp_prop class.
8220 (search_for_addr_array, vrp_initialize): Likewise.
8221 (vrp_finalize): Likewise. Revamp to avoid direct access to
8222 vr_value, values_propagated, etc.
8223 (check_array_bounds): Extract vrp_prop class instance pointer from
8224 walk info structure. Use it to call member functions.
8225 (check_all_array_refs): Make a member function within vrp_prop class.
8226 Smuggle class instance pointer via walk info structure.
8227 (x_vr_values): New local static.
8228 (vrp_valueize): Use x_vr_values to get class instance.
8229 (vr_valueize_1): Likewise.
8230 (class vrp_folder): Add vr_values data member. Add various member
8231 functions as well as member functions that delegate to vr_values.
8232 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
8233 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
8234 class instance from vr_values. Use it to call member functions.
8235 (vrp_dom_walker): Add vr_values data member.
8236 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
8237 instance via x_vr_values.
8238 (identify_jump_threads): Accept vr_values as argument. Store
8239 it into the walker structure.
8240 (evrp_dom_walker): Add vr_values class data member. Add various
8242 (evrp_dom_walker::try_find_new_range): Use vr_values data
8243 member to access the memory allocator.
8244 (evrp_dom_walker::before_dom_children): Store vr_values class
8245 instance into the vrp_folder class.
8246 (evrp_dom_walker::push_value_range): Rework to avoid direct
8247 access to num_vr_values and vr_value.
8248 (evrp_dom_walker::pop_value_range): Likewise.
8249 (execute_early_vrp): Remove call to vrp_initialize_lattice.
8250 Use vr_values to get to dump_all_value_ranges member function.
8251 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
8252 and simplify_cond_using_ranges_2 via vrp_prop class instance.
8253 Pass vr_values class instance down to identify_jump_threads.
8254 Remove call to vrp_free_lattice.
8255 (debug_all_value_ranges): Remove.
8257 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
8258 (vrp_folder): Likewise.
8260 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
8261 Get it from the existing bitmap instead.
8262 (vrp_intersect_ranges_1): Likewise.
8264 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8266 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8268 (store_immediate_info::store_immediate_info): Add bitnotp argument,
8269 set bit_not_p to it.
8270 (imm_store_chain_info::coalesce_immediate_stores): Break group
8271 if bit_not_p is different.
8272 (count_multiple_uses, split_group,
8273 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
8274 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
8275 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
8276 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
8277 to store_immediate_info ctor.
8279 2017-11-09 Jim Wilson <jimw@sifive.com>
8281 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
8282 (scan_prog_file): Likewise.
8284 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8286 * bb-reorder.c (max_entry_frequency): Remove.
8287 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
8288 connect_traces, push_to_next_round_p): Remove prototypes.
8289 (find_traces_1_round): Use counts only.
8290 (push_to_next_round_p): Likewise.
8291 (find_traces): Likewise.
8292 (rotate_loop): Likewise.
8293 (find_traces_1_round): Likewise.
8294 (connect_traces): Likewise.
8295 (edge_order): Likewise.
8297 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
8299 * config/arm/arm.c (output_return_instruction): Add comments to
8300 indicate requirement for cmse_nonsecure_entry return to account
8301 for the size of clearing instruction output here.
8302 (thumb_exit): Likewise.
8303 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
8304 return in hardfloat mode.
8306 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
8308 * config/rs6000/rs6000.c (machine_function): Add a bool,
8309 "toc_is_wrapped_separately".
8310 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
8311 if it wasn't explicitly set or unset, we are optimizing for speed, and
8312 doing separate shrink-wrapping.
8313 (rs6000_get_separate_components): Enable the TOC component if
8314 saving the TOC register in the prologue.
8315 (rs6000_components_for_bb): Handle the TOC component.
8316 (rs6000_emit_prologue_components): Store the TOC register where needed.
8317 (rs6000_set_handled_components): Mark TOC as handled, if handled.
8318 (rs6000_emit_prologue): Don't save the TOC if that is already done.
8320 2017-11-09 Martin Jambor <mjambor@suse.cz>
8322 * ipa-param-manipulation.c: New file.
8323 * ipa-param-manipulation.h: Likewise.
8324 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
8325 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
8326 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
8327 (ipa_parm_adjustment): Likewise.
8328 (ipa_parm_adjustment_vec): Likewise.
8329 (ipa_get_vector_of_formal_parms): Moved declaration to
8330 ipa-param-manipulation.h.
8331 (ipa_get_vector_of_formal_parm_types): Likewise.
8332 (ipa_modify_formal_parameters): Likewise.
8333 (ipa_modify_call_arguments): Likewise.
8334 (ipa_combine_adjustments): Likewise.
8335 (ipa_dump_param_adjustments): Likewise.
8336 (ipa_modify_expr): Likewise.
8337 (ipa_get_adjustment_candidate): Likewise.
8338 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
8339 ipa-param-manipulation.c.
8340 (ipa_get_vector_of_formal_parm_types): Likewise.
8341 (ipa_modify_formal_parameters): Likewise.
8342 (ipa_modify_call_arguments): Likewise.
8343 (ipa_modify_expr): Likewise.
8344 (get_ssa_base_param): Likewise.
8345 (ipa_get_adjustment_candidate): Likewise.
8346 (index_in_adjustments_multiple_times_p): Likewise.
8347 (ipa_combine_adjustments): Likewise.
8348 (ipa_dump_param_adjustments): Likewise.
8349 * tree-sra.c: Also include ipa-param-manipulation.h
8350 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
8353 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8354 Alan Hayward <alan.hayward@arm.com>
8355 David Sherwood <david.sherwood@arm.com>
8357 * doc/sourcebuild.texi (vect_masked_store): Document.
8359 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8360 Alan Hayward <alan.hayward@arm.com>
8361 David Sherwood <david.sherwood@arm.com>
8363 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
8365 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8366 Alan Hayward <alan.hayward@arm.com>
8367 David Sherwood <david.sherwood@arm.com>
8369 * doc/sourcebuild.texi (vect_variable_length): Document.
8371 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8372 Alan Hayward <alan.hayward@arm.com>
8373 David Sherwood <david.sherwood@arm.com>
8375 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
8377 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8378 Alan Hayward <alan.hayward@arm.com>
8379 David Sherwood <david.sherwood@arm.com>
8381 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
8383 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8384 Alan Hayward <alan.hayward@arm.com>
8385 David Sherwood <david.sherwood@arm.com>
8387 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
8388 previously undocumented selectors.
8389 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
8391 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8393 * doc/rtl.texi (const_vector): Say that elements can be
8394 const_wide_ints too.
8395 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
8396 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
8397 (gen_vec_duplicate): Use it instead of CONSTANT_P.
8398 * optabs.c (expand_vector_broadcast): Likewise.
8400 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8401 Alan Hayward <alan.hayward@arm.com>
8402 David Sherwood <david.sherwood@arm.com>
8404 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
8405 scaled index even if the unscaled address was invalid.
8406 Don't increase the complexity of using a scale in that case.
8408 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8409 Alan Hayward <alan.hayward@arm.com>
8410 David Sherwood <david.sherwood@arm.com>
8412 * doc/rtl.texi: Rewrite the subreg rules so that they partition
8413 the inner register into REGMODE_NATURAL_SIZE bytes rather than
8414 UNITS_PER_WORD bytes.
8415 * emit-rtl.c (validate_subreg): Divide subregs into blocks
8416 based on REGMODE_NATURAL_SIZE of the inner mode.
8417 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
8418 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
8419 * expmed.c (lowpart_bit_field_p): Divide the value up into
8420 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
8421 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
8422 whether something is likely to occupy more than one register.
8424 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8427 * ipa-inline-transform.c (update_noncloned_frequencies): Use
8428 profile_count::adjust_for_ipa_scaling.
8429 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
8430 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
8432 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
8434 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8436 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
8437 (split_group): Add total_orig and total_new arguments, estimate the
8438 number of statements related to the store group without store merging
8439 and with store merging.
8440 (imm_store_chain_info::output_merged_store): Adjust split_group
8441 callers, punt if estimated number of statements with store merging
8442 is not smaller than estimated number of statements without it.
8444 (handled_load): Remove has_single_use checks.
8445 (pass_store_merging::process_store): Likewise.
8447 2017-11-09 Richard Biener <rguenther@suse.de>
8449 PR tree-optimization/82902
8450 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
8452 2017-11-09 Martin Liska <mliska@suse.cz>
8455 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
8458 2017-11-09 Martin Liska <mliska@suse.cz>
8460 PR tree-optimization/82669
8461 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
8463 2017-11-09 Martin Liska <mliska@suse.cz>
8465 PR gcov-profile/48463
8466 * coverage.c (coverage_begin_function): Output also end locus
8467 of a function and information whether the function is
8469 * gcov-dump.c (tag_function): Parse and print the information.
8470 * gcov.c (INCLUDE_MAP): Add include.
8471 (INCLUDE_SET): Likewise.
8472 (struct line_info): Move earlier in the source file because
8473 of vector<line_info> in function_info structure.
8474 (line_info::line_info): Likewise.
8475 (line_info::has_block): Likewise.
8476 (struct source_info): Add new member index.
8477 (source_info::get_functions_at_location): New function.
8478 (function_info::group_line_p): New function.
8479 (output_intermediate_line): New function.
8480 (output_intermediate_file): Use the mentioned function.
8481 (struct function_start): New.
8482 (struct function_start_pair_hash): Likewise.
8483 (process_file): Add code that identifies group functions.
8484 Assign lines either to global or function scope.
8485 (generate_results): Skip artificial functions.
8486 (find_source): Assign index for each source file.
8487 (read_graph_file): Read new flag artificial and end_line.
8488 (add_line_counts): Assign it either to global of function scope.
8489 (accumulate_line_counts): Isolate core of the function to
8490 accumulate_line_info and call it for both function and global
8492 (accumulate_line_info): New function.
8493 (output_line_beginning): Fix GNU coding style.
8494 (print_source_line): New function.
8495 (output_line_details): Likewise.
8496 (output_function_details): Likewise.
8497 (output_lines): Iterate both source (global) scope and function
8499 (struct function_line_start_cmp): New class.
8500 * doc/gcov.texi: Reflect changes in documentation.
8502 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8505 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
8506 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
8507 and similarly for not instead of neg.
8509 2017-11-08 Andi Kleen <ak@linux.intel.com>
8511 * config/i386/i386.opt: Add -mforce-indirect-call.
8512 * config/i386/predicates.md: Check for flag_force_indirect_call.
8513 * doc/invoke.texi: Document -mforce-indirect-call
8515 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
8517 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
8519 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
8520 config/riscv/riscv.c (predict.h): New include.
8521 (riscv_slow_unaligned_access_p): No longer static.
8522 (riscv_block_move_straight): Add require.
8523 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8525 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8528 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
8529 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
8530 nonimmediate_operand predicate for operand 1 instead of
8533 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8535 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
8537 * config/aarch64/constraints.md (Uml): New constraint.
8538 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
8541 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8543 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
8544 of two vec_duplicates into a vec_concat.
8546 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8548 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8549 Simplify vec_merge of vec_duplicate and vec_concat.
8550 * config/aarch64/constraints.md (Utq): New constraint.
8551 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
8554 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8556 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8557 Simplify vec_merge of vec_duplicate and const_vector.
8558 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
8560 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
8561 mode iterator. Update predicate on operand 1 to
8562 handle non-const_vec constants. Delete constraints.
8563 (*aarch64_combinez_be<mode>): Likewise for operand 2.
8565 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8567 PR tree-optimization/78821
8568 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
8570 (store_operand_info::store_operand_info): Initialize it to false.
8571 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
8572 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
8573 store in the group, and if chain_info is non-NULL, to ignore altogether
8575 (compatible_load_p): Fail if bit_not_p does not match.
8576 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
8577 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
8578 (pass_store_merging::process_store): Adjust
8579 terminate_all_aliasing_chains calls to pass NULL in all current spots,
8580 call terminate_all_aliasing_chains newly when adding a store into
8581 a chain with non-NULL chain_info.
8583 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8585 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
8587 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8589 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8591 (aarch64_layout_frame): Initialise emit_frame_chain.
8592 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
8593 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
8595 2017-11-08 Martin Liska <mliska@suse.cz>
8597 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
8598 of gimple_call_internal_p.
8600 2017-11-07 Tom de Vries <tom@codesourcery.com>
8602 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
8604 2017-11-07 Tom de Vries <tom@codesourcery.com>
8606 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
8609 2017-11-08 Martin Liska <mliska@suse.cz>
8612 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
8614 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8616 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
8618 2017-11-07 Jakub Jelinek <jakub@redhat.com>
8621 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
8622 (*cmp<mode>_ccz_1): New insn with $k alternative.
8625 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
8626 RTX_COMM_COMPARE as commutative as well.
8627 (ix86_binary_operator_ok): Formatting fix.
8628 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
8629 *<code><mode>3<mask_name><round_saeonly_name>,
8630 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
8631 *<s>mul<mode>3_highpart<mask_name>,
8632 *vec_widen_umult_even_v16si<mask_name>,
8633 *vec_widen_umult_even_v8si<mask_name>,
8634 *vec_widen_umult_even_v4si<mask_name>,
8635 *vec_widen_smult_even_v16si<mask_name>,
8636 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
8637 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
8638 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
8639 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
8640 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
8641 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
8642 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
8643 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
8644 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
8645 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
8646 ix86_binary_operator_ok. Formatting fixes.
8647 (*<plusminus_insn><mode>3<mask_name><round_name>,
8648 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
8651 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
8653 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
8654 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
8655 eq and ne if TARGET_ISEL.
8656 (cmp): New code_iterator.
8657 (UNS, UNSU_, UNSIK): New code_attrs.
8658 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
8659 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
8661 ("eq<mode>3"): ... this.
8662 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
8664 ("ne<mode>3"): ... this.
8666 2017-11-07 Julia Koval <julia.koval@intel.com>
8669 * common/config/i386/i386-common.c
8670 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
8671 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
8672 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
8673 * config/i386/i386.opt: Ditto.
8674 * config/i386/i386.c (ix86_target_string): Ditto.
8675 (ix86_option_override_internal): Ditto.
8676 (ix86_init_mpx_builtins): Move MPX to args2.
8677 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
8678 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8679 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8680 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8681 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8682 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8683 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
8685 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
8688 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
8689 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
8690 (zero-extendsidi peephole2): Remove peephole.
8692 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8695 * stor-layout.c: Include attribs.h.
8696 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
8697 explicit lookup of "aligned" attribute.
8699 2017-11-07 Andrew Waterman <andrew@sifive.com>
8701 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
8702 (riscv_expand_block_move): Likewise.
8703 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
8705 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
8706 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
8707 * config/riscv/riscv.c (riscv_block_move_straight): New function.
8708 (riscv_adjust_block_mem): Likewise.
8709 (riscv_block_move_loop): Likewise.
8710 (riscv_expand_block_move): Likewise.
8711 * config/riscv/riscv.md (movmemsi): New pattern.
8713 2017-11-07 Michael Clark <michaeljclark@mac.com>
8715 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
8716 (MUSL_DYNAMIC_LINKER): Likewise.
8718 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
8720 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
8721 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
8724 2017-11-07 Richard Biener <rguenther@suse.de>
8726 * match.pd: Fix build.
8728 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
8729 Jackson Woodruff <jackson.woodruff@arm.com>
8731 PR tree-optimization/71026
8732 * match.pd: Canonicalize negate in division.
8734 2017-11-07 Sudakshina Das <sudi.das@arm.com>
8737 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
8739 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8741 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
8742 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
8744 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8746 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
8747 non-scalar integral types.
8748 * match.pd (negate_expr_p): Handle MINUS_EXPR.
8749 (-(A-B), -(~A)): New transformations.
8751 2017-11-07 Tom de Vries <tom@codesourcery.com>
8753 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
8754 semicolon after "do {} while (0)".
8755 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8756 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8757 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8758 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8759 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8761 2017-11-07 Tom de Vries <tom@codesourcery.com>
8763 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
8764 after "do {} while (0)".
8765 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8766 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8767 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8768 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8769 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8771 2017-11-07 Tom de Vries <tom@codesourcery.com>
8774 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
8776 (arm_rtx_costs_internal): Add missing semicolon after
8777 HANDLE_NARROW_SHIFT_ARITH call.
8779 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8781 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8782 disable isel if it was not set explicitly.
8784 2017-11-06 James Bowman <james.bowman@ftdichip.com>
8786 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
8787 (add_type_attribute) likewise.
8789 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
8791 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
8792 of struct ix86_frame.
8793 (ix86_initial_elimination_offset): Likewise.
8794 (ix86_expand_split_stack_prologue): Likewise.
8796 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
8798 * tree-vrp.h (enum value_range_type): Update stale comment.
8800 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8801 Alan Hayward <alan.hayward@arm.com>
8802 David Sherwood <david.sherwood@arm.com>
8804 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
8805 (aarch64_expand_vec_perm_const): Take the number of units too.
8806 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8807 (aarch64_expand_vec_perm_const): Likewise.
8808 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8809 (vec_perm<mode>): Update accordingly.
8811 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8812 Alan Hayward <alan.hayward@arm.com>
8813 David Sherwood <david.sherwood@arm.com>
8815 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8816 Take the number of units too.
8817 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8818 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8819 but check for a vector mode before rather than after the call.
8820 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8821 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8822 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8823 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8824 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8825 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8826 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8827 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8828 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8829 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8830 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8831 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8832 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8833 (aarch64_sqdmull2_n<mode>): Update accordingly.
8835 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8836 Alan Hayward <alan.hayward@arm.com>
8837 David Sherwood <david.sherwood@arm.com>
8839 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8840 the number of units too.
8841 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8842 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8843 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
8844 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
8845 (vec_store_lanesxi<mode>): Update accordingly.
8847 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8848 Alan Hayward <alan.hayward@arm.com>
8849 David Sherwood <david.sherwood@arm.com>
8851 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8852 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8853 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8854 of units rather than the mode.
8855 * config/aarch64/iterators.md (nunits): New mode attribute.
8856 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8857 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8858 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8859 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8860 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8861 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8862 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8863 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8864 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8865 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8866 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8867 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8868 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8869 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8870 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8871 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8872 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8873 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8874 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8875 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8876 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8877 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8878 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8879 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8880 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8881 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8882 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8883 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8884 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8885 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8886 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8887 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8888 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8889 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8890 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8891 (aarch64_simd_vec_setv2di): Likewise.
8893 2017-11-06 Carl Love <cel@us.ibm.com>
8895 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8897 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8898 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8900 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8902 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8903 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8904 Add power 8 macro expansions.
8905 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8906 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8907 power 8 instructions. (VSX_XXBR): Add iterator.
8909 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8911 * config/arm/arm.md (predicable_short_it): Change default to "no",
8912 improve documentation, remove uses that are identical to the default.
8913 (enabled_for_depr_it): Rename to enabled_for_short_it.
8914 * gcc/config/arm/arm-fixed.md (predicable_short_it):
8915 Remove default uses.
8916 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
8917 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
8918 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
8919 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
8921 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
8924 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
8925 float128 helper macros, which are no longer used after deleting
8926 the old 'q' built-in functions, and moving the round to odd
8927 built-in functions to being special built-in functions.
8928 (BU_FLOAT128_2): Likewise.
8929 (BU_FLOAT128_1_HW): Likewise.
8930 (BU_FLOAT128_2_HW): Likewise.
8931 (BU_FLOAT128_3_HW): Likewise.
8932 (FABSQ): Delete old 'q' built-in functions.
8933 (COPYSIGNQ): Likewise.
8934 (SQRTF128_ODD): Move round to odd built-in functions to be
8935 special built-in functions, so that we can handle
8936 -mabi=ieeelongdouble.
8937 (TRUNCF128_ODD): Likewise.
8938 (ADDF128_ODD): Likewise.
8939 (SUBF128_ODD): Likewise.
8940 (MULF128_ODD): Likewise.
8941 (DIVF128_ODD): Likewise.
8942 (FMAF128_ODD): Likewise.
8943 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
8944 built-in names to 'f128'.
8945 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
8946 old 'q' built-in functions, as the machine independent code for
8947 'f128' built-in functions handles this.
8948 (rs6000_expand_builtin): Add expansion for float128 round to odd
8949 functions, keying off on -mabi=ieeelongdouble of whether to use
8950 the KFmode or TFmode variant.
8951 (rs6000_init_builtins): Initialize the _Float128 round to odd
8953 * doc/extend.texi (PowerPC Built-in Functions): Document the old
8954 _Float128 'q' built-in functions are now mapped into the new
8955 'f128' built-in functions.
8957 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
8959 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
8960 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
8962 2017-11-06 David Malcolm <dmalcolm@redhat.com>
8965 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
8966 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
8967 * toplev.c: Include "ipa-fnsummary.h".
8968 (toplev::finalize): Call ipa_fnsummary_c_finalize.
8970 2017-11-06 Jakub Jelinek <jakub@redhat.com>
8972 PR tree-optimization/82838
8973 * gimple-ssa-store-merging.c
8974 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
8975 on a separate gimple_seq which is then appended to seq.
8977 2017-11-06 Jeff Law <law@redhat.com>
8980 * config/i386/i386.c (PROBE_INTERVAL): Remove.
8981 (get_probe_interval): New functions.
8982 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
8983 (ix86_adjust_stack_and_probe): Likewise.
8984 (output_adjust_stack_and_probe): Likewise.
8985 (ix86_emit_probe_stack_range): Likewise.
8986 (ix86_expand_prologue): Likewise.
8988 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8990 PR tree-optimization/82816
8991 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
8992 if the modes of the two types are the same.
8993 (convert_plusminus_to_widen): Likewise.
8995 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8997 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
8999 (usadv16qi): New define_expand.
9000 (usadv8hi): New define_expand.
9002 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
9005 * ipa-inline-transform.c (update_noncloned_frequencies): Always
9007 (inline_transform): Likewise.
9008 * predict.c (counts_to_freqs): Remove useless conditional.
9009 * profile-count.h (profile_count::apply_scale): Move sanity check.
9010 * tree-inline.c (copy_bb): Always scale.
9011 (copy_cfg_body): Likewise.
9013 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
9016 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
9018 (*cmp_ite0): Add enabled_for_depr_it attribute.
9019 (*cmp_ite1): Likewise.
9021 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
9023 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
9026 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
9028 * tree-vrp.c (vrp_int_const_binop): Return true on success and
9029 return the value by pointer.
9030 (extract_range_from_multiplicative_op_1): Update accordingly.
9031 Return as soon as an operation fails.
9033 2017-11-05 Tom de Vries <tom@codesourcery.com>
9036 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
9037 (DEF_SANITIZER_BUILTIN): ... here.
9038 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
9039 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
9041 2017-11-05 Tom de Vries <tom@codesourcery.com>
9044 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
9046 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
9047 ASM_OUTPUT_BEFORE_CASE_LABEL call.
9048 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
9050 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9051 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9052 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9054 2017-11-05 Tom de Vries <tom@codesourcery.com>
9057 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
9060 2017-11-04 Michael Clark <michaeljclark@mac.com>
9062 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
9063 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
9065 (*addsi3_extended): Likewise.
9066 (*addsi3_extended2): Likewise.
9067 (<optab>si3): Likewise.
9068 (<optab>di3): Likewise.
9069 (<optab><mode>3): Likewise.
9070 (<*optabe>si3_internal): Likewise.
9071 (zero_extendqi<SUPERQI:mode>2): Likewise.
9072 (*add<mode>hi3): Likewise.
9073 (*xor<mode>hi3): Likewise.
9074 (<optab>di3): Likewise.
9075 (*<optab>si3_extend): Likewise.
9076 (*sge<u>_<X:mode><GPR:mode>): Likewise.
9077 (*slt<u>_<X:mode><GPR:mode>): Likewise.
9078 (*sle<u>_<X:mode><GPR:mode>): Likewise.
9080 2017-11-04 Andrew Waterman <andrew@sifive.com>
9082 * config/riscv/riscv.c (riscv_option_override): Conditionally set
9083 TARGET_STRICT_ALIGN based upon -mtune argument.
9085 2017-11-04 Andrew Waterman <andrew@sifive.com>
9087 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
9089 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
9091 * config/i386/i386.c (choose_basereg): Use optional scratch
9092 register and add assertion.
9093 (x86_emit_outlined_ms2sysv_save): Use scratch register when
9094 needed, and don't allocate stack.
9095 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
9096 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
9097 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
9099 2017-11-03 Jeff Law <law@redhat.com>
9101 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
9102 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
9103 to probe at the start of a noreturn function.
9105 2017-11-03 Jakub Jelinek <jakub@redhat.com>
9107 PR tree-optimization/78821
9108 * gimple-ssa-store-merging.c: Update the file comment.
9109 (MAX_STORE_ALIAS_CHECKS): Define.
9110 (struct store_operand_info): New type.
9111 (store_operand_info::store_operand_info): New constructor.
9112 (struct store_immediate_info): Add rhs_code and ops data members.
9113 (store_immediate_info::store_immediate_info): Add rhscode, op0r
9114 and op1r arguments to the ctor, initialize corresponding data members.
9115 (struct merged_store_group): Add load_align_base and load_align
9117 (merged_store_group::merged_store_group): Initialize them.
9118 (merged_store_group::do_merge): Update them.
9119 (merged_store_group::apply_stores): Pick the constant for
9120 encode_tree_to_bitpos from one of the two operands, or skip
9121 encode_tree_to_bitpos if neither operand is a constant.
9122 (class pass_store_merging): Add process_store method decl. Remove
9123 bool argument from terminate_all_aliasing_chains method decl.
9124 (pass_store_merging::terminate_all_aliasing_chains): Remove
9125 var_offset_p argument and corresponding handling.
9126 (stmts_may_clobber_ref_p): New function.
9127 (compatible_load_p): New function.
9128 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
9129 if there is overlap and rhs_code is not INTEGER_CST. For
9130 non-overlapping stores terminate group if rhs is not mergeable.
9131 (get_alias_type_for_stmts): Change first argument from
9132 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
9133 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
9134 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
9136 (get_location_for_stmts): Change first argument from
9137 auto_vec<gimple *> & to vec<gimple *> &.
9138 (struct split_store): Remove orig_stmts data member, add orig_stores.
9139 (split_store::split_store): Create orig_stores rather than orig_stmts.
9140 (find_constituent_stmts): Renamed to ...
9141 (find_constituent_stores): ... this. Change second argument from
9142 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
9143 to info structures rather than the statements.
9144 (split_group): Rename ALLOW_UNALIGNED argument to
9145 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
9146 it. Adjust find_constituent_stores caller.
9147 (imm_store_chain_info::output_merged_store): Handle rhs_code other
9148 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
9149 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
9150 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
9151 (mem_valid_for_store_merging): New function.
9152 (handled_load): New function.
9153 (pass_store_merging::process_store): New method.
9154 (pass_store_merging::execute): Use process_store method. Adjust
9155 terminate_all_aliasing_chains caller.
9157 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9159 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
9160 Return true for more constants, symbols and label references.
9161 (aarch64_valid_floating_const): Remove unused function.
9163 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9166 * config/aarch64/aarch64.c (aarch64_layout_frame):
9167 Undo forcing of LR at bottom of frame.
9169 2017-11-03 Jeff Law <law@redhat.com>
9172 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
9173 for int_registers_saved.
9175 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
9176 extracted from tree-ssa-dom.c.
9177 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
9178 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
9179 (record_equivalences_from_incoming_edge): Add additional argument
9180 to single_pred_edge_ignoring_loop_edges call.
9181 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
9182 (uncprop_dom_walker::before_dom_children): Add additional argument
9183 to single_pred_edge_ignoring_loop_edges call.
9184 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
9185 single_pred_edge_ignoring_loop_edges rather than open coding.
9186 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
9188 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
9190 * match.pd (-(-A)): Rewrite.
9192 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9194 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
9195 (rs6000_emit_int_cmove): New declaration.
9196 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
9197 (rs6000_emit_sISEL): Delete.
9198 (rs6000_emit_int_cmove): Make non-static.
9199 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
9200 instead of rs6000_emit_sISEL.
9202 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
9204 * asan.c (create_cond_insert_point): Maintain profile.
9205 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
9207 * basic-block.h (struct basic_block_def): Remove frequency.
9208 (EDGE_FREQUENCY): Use to_frequency
9209 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
9211 (find_traces): Update to use to_frequency.
9212 (find_traces_1_round): Likewise; use only IPA counts.
9213 (bb_to_key): Likewise.
9214 (connect_traces): Use IPA counts only.
9215 (copy_bb_p): Update to use to_frequency.
9216 (fix_up_crossing_landing_pad): Likewise.
9217 (sanitize_hot_paths): Likewise.
9218 * bt-load.c (basic_block_freq): Likewise.
9219 * cfg.c (init_flow): Set count_max to uninitialized.
9220 (check_bb_profile): Remove frequencies; check counts.
9221 (dump_bb_info): Do not dump frequencies.
9222 (update_bb_profile_for_threading): Update counts only.
9223 (scale_bbs_frequencies_int): Likewise.
9224 (MAX_SAFE_MULTIPLIER): Remove.
9225 (scale_bbs_frequencies_gcov_type): Update counts only.
9226 (scale_bbs_frequencies_profile_count): Update counts only.
9227 (scale_bbs_frequencies): Update counts only.
9228 * cfg.h (struct control_flow_graph): Add count-max.
9229 (update_bb_profile_for_threading): Update prototype.
9230 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
9231 (find_many_sub_basic_blocks): Likewise.
9232 * cfgcleanup.c (try_forward_edges): Likewise.
9233 (try_crossjump_to_edge): Likewise.
9234 * cfgexpand.c (expand_gimple_cond): Likewise.
9235 (expand_gimple_tailcall): Likewise.
9236 (construct_init_block): Likewise.
9237 (construct_exit_block): Likewise.
9238 * cfghooks.c (verify_flow_info): Check consistency of counts.
9239 (dump_bb_for_graph): Do not dump frequencies.
9240 (split_block_1): Do not update frequencies.
9241 (split_edge): Do not update frequencies.
9242 (make_forwarder_block): Do not update frequencies.
9243 (duplicate_block): Do not update frequencies.
9244 (account_profile_record): Do not update frequencies.
9245 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
9246 for global heuristics.
9247 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
9248 (expected_loop_iterations_unbounded): Use counts only.
9249 * cfgloopmanip.c (scale_loop_profile): Simplify.
9250 (create_empty_loop_on_edge): Simplify
9252 (duplicate_loop_to_header_edge): Simplify
9253 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
9254 (update_br_prob_note): Take care of removing note when profile
9256 (relink_block_chain): Do not dump frequency.
9257 (rtl_account_profile_record): Use to_frequency.
9258 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
9259 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
9260 (cgraph_update_edges_for_call_stmt_node): Likewise.
9261 (cgraph_edge::verify_count_and_frequency): Update.
9262 (cgraph_node::verify_node): Temporarily disable frequency verification.
9263 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
9264 to_cgraph_frequency.
9265 (cgraph_edge::rebuild_edges): Convert to ipa counts.
9266 * cgraphunit.c (init_lowered_empty_function): Do not initialize
9268 (cgraph_node::expand_thunk): Update profile.
9269 * except.c (dw2_build_landing_pads): Do not update frequency.
9270 * final.c (compute_alignments): Use to_frequency.
9271 (dump_basic_block_info): Do not dump frequency.
9272 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
9273 (dump_gimple_bb_header): Do not dump frequency.
9274 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
9276 * gimple-streamer-in.c (input_bb): Do not stream frequency.
9277 * gimple-streamer-out.c (output_bb): Do not stream frequency.
9278 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
9279 (init_before_recovery): Do not update frequency.
9280 (sched_create_recovery_edges): Do not update frequency.
9281 * hsa-gen.c (convert_switch_statements): Do not update frequency.
9282 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
9283 (ipa_cp_c_finalize): Set max_count to uninitialized.
9284 * ipa-fnsummary.c (get_minimal_bb): Use counts.
9285 (param_change_prob): Use counts.
9286 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
9288 * ipa-split.c (consider_split): Use to_frequency.
9289 (split_function): Use to_frequency.
9290 * ira-build.c (loop_compare_func): Likewise.
9291 (mark_loops_for_removal): Likewise.
9292 (mark_all_loops_for_removal): Likewise.
9293 * loop-doloop.c (doloop_modify): Do not update frequency.
9294 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
9296 * lto-streamer-in.c (input_function): Update count_max.
9297 * omp-expand.c (expand_omp_taskreg): Update count_max.
9298 * omp-simd-clone.c (simd_clone_adjust): Update profile.
9299 * predict.c (maybe_hot_frequency_p): Use to_frequency.
9300 (maybe_hot_count_p): Use ipa counts only.
9301 (maybe_hot_bb_p): Simplify.
9302 (maybe_hot_edge_p): Simplify.
9303 (probably_never_executed): Do not take frequency argument.
9304 (probably_never_executed_bb_p): Do not pass frequency.
9305 (probably_never_executed_edge_p): Likewise.
9306 (combine_predictions_for_bb): Check that profile is nonzero.
9307 (propagate_freq): Do not set frequency.
9308 (drop_profile): Simplify.
9309 (counts_to_freqs): Simplify.
9310 (expensive_function_p): Use to_frequency.
9311 (propagate_unlikely_bbs_forward): Simplify.
9312 (determine_unlikely_bbs): Simplify.
9313 (estimate_bb_frequencies): Add hack to silence graphite issues.
9314 (compute_function_frequency): Use ipa counts.
9315 (pass_profile::execute): Update.
9316 (rebuild_frequencies): Use counts only.
9317 (force_edge_cold): Use counts only.
9318 * profile-count.c (profile_count::dump): Dump new count types.
9319 (profile_count::differs_from_p): Check compatiblity.
9320 (profile_count::to_frequency): New function.
9321 (profile_count::to_cgraph_frequency): New function.
9322 * profile-count.h (struct function): Declare.
9323 (enum profile_quality): Add profile_guessed_local and
9324 profile_guessed_global0.
9325 (class profile_proability): Decrease number of bits to 29;
9326 update from_reg_br_prob_note and to_reg_br_prob_note.
9327 (class profile_count: Update comment; decrease number of bits
9328 to 61. Check compatibility.
9329 (profile_count::compatible_p): New private member function.
9330 (profile_count::ipa_p): New member function.
9331 (profile_count::operator<): Handle global zero correctly.
9332 (profile_count::operator>): Handle global zero correctly.
9333 (profile_count::operator<=): Handle global zero correctly.
9334 (profile_count::operator>=): Handle global zero correctly.
9335 (profile_count::nonzero_p): New member function.
9336 (profile_count::force_nonzero): New member function.
9337 (profile_count::max): New member function.
9338 (profile_count::apply_scale): Handle IPA scalling.
9339 (profile_count::guessed_local): New member function.
9340 (profile_count::global0): New member function.
9341 (profile_count::ipa): New member function.
9342 (profile_count::to_frequency): Declare.
9343 (profile_count::to_cgraph_frequency): Declare.
9344 * profile.c (OVERLAP_BASE): Delete.
9345 (compute_frequency_overlap): Delete.
9346 (compute_branch_probabilities): Do not use compute_frequency_overlap.
9347 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
9348 * sched-ebb.c (rank): Use counts only.
9349 * shrink-wrap.c (handle_simple_exit): Use counts only.
9350 (try_shrink_wrapping): Use counts only.
9351 (place_prologue_for_one_component): Use counts only.
9352 * tracer.c (find_best_predecessor): Use to_frequency.
9353 (find_trace): Use to_frequency.
9354 (tail_duplicate): Use to_frequency.
9355 * trans-mem.c (expand_transaction): Do not update frequency.
9356 * tree-call-cdce.c: Do not update frequency.
9357 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
9358 (gimple_merge_blocks): Likewise.
9359 (gimple_split_edge): Likewise.
9360 (gimple_duplicate_sese_region): Likewise.
9361 (gimple_duplicate_sese_tail): Likewise.
9362 (move_sese_region_to_fn): Likewise.
9363 (gimple_account_profile_record): Likewise.
9364 (insert_cond_bb): Likewise.
9365 * tree-complex.c (expand_complex_div_wide): Likewise.
9366 * tree-eh.c (lower_resx): Update profile.
9367 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
9369 (initialize_cfun): Do not initialize frequencies
9370 (freqs_to_counts): Delete.
9371 (copy_cfg_body): Ignore count parameter.
9372 (copy_body): Update.
9373 (expand_call_inline): Update count_max.
9374 (optimize_inline_calls): Update count_max.
9375 (tree_function_versioning): Update count_max.
9376 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
9377 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
9379 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
9380 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
9381 (try_peel_loop): Likewise.
9382 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
9384 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
9385 (tree_transform_and_unroll_loop): Do not use frequencies
9386 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
9387 Use reliable prediction only.
9388 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
9389 * tree-ssa-sink.c (select_best_block): Use to_frequency.
9390 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
9391 probability scaling.
9392 * tree-ssa-threadupdate.c (create_block_for_threading): Do
9393 not update frequency
9394 (any_remaining_duplicated_blocks): Likewise.
9395 (update_profile): Likewise.
9396 (estimated_freqs_path): Delete.
9397 (freqs_to_counts_path): Delete.
9398 (clear_counts_path): Delete.
9399 (ssa_fix_duplicate_block_edges): Likewise.
9400 (duplicate_thread_path): Likewise.
9401 * tree-switch-conversion.c (gen_inbound_check): Use counts.
9402 * tree-tailcall.c (decrease_profile): Do not update frequency.
9403 (eliminate_tail_call): Likewise.
9404 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
9405 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
9406 (optimize_mask_stores): Likewise.
9407 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9408 * ubsan.c (ubsan_expand_null_ifn): Update profile.
9409 (ubsan_expand_ptr_ifn): Update profile.
9410 * value-prof.c (gimple_ic): Simplify.
9411 * value-prof.h (gimple_ic): Update prototype.
9412 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
9413 * ipa-inline.c (compute_uninlined_call_time): Be sure that
9415 (want_inline_self_recursive_call_p): Likewise.
9416 (resolve_noninline_speculation): Only cummulate defined counts.
9417 (inline_small_functions): Use nonzero_p.
9418 (ipa_inline): Do not access freed node.
9420 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9422 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9423 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
9425 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
9427 * config/riscv/riscv.c (riscv_legitimize_move): Handle
9428 non-legitimate address.
9430 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9432 * config/rs6000/rs6000.md (*lt0_disi): Delete.
9433 (*lt0_<mode>di, *lt0_<mode>si): New.
9435 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9437 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
9438 TARGET_PAIRED_FLOAT.
9440 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
9441 Jim Wilson <jim.wilson@linaro.org>
9443 * config/aarch64/aarch64-cores.def (saphira): New CPU.
9444 * config/aarch64/aarch64-tune.md: Regenerated.
9445 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
9446 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
9448 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
9450 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
9451 (arc_expand_prologue): Restore blink for millicode.
9452 * config/arc/linux.h (LINK_EH_SPEC): Defined.
9454 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
9457 * config/i386/i386.c (ix86_vector_duplicate_value): Use
9458 gen_vec_duplicate after forcing the scalar into a register.
9460 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
9462 * combine (try_combine): Print the insns input to try_combine to the
9465 2017-11-02 Steve Ellcey <sellcey@cavium.com>
9468 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
9469 Remove second argument from aarch64_process_target_attr call.
9470 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
9472 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
9474 (aarch64_handle_attr_arch): Remove second argument.
9475 (aarch64_handle_attr_cpu): Ditto.
9476 (aarch64_handle_attr_tune): Ditto.
9477 (aarch64_handle_attr_isa_flags): Ditto.
9478 (aarch64_process_one_target_attr): Ditto.
9479 (aarch64_process_target_attr): Ditto.
9480 (aarch64_option_valid_attribute_p): Remove second argument.
9481 on aarch64_process_target_attr call.
9483 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9485 * diagnostic.c: Include "selftest-diagnostic.h".
9486 (selftest::assert_location_text): New function.
9487 (selftest::test_diagnostic_get_location_text): New function.
9488 (selftest::diagnostic_c_tests): Call it.
9490 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9492 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
9493 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
9494 (class selftest::test_diagnostic_context): Move to...
9495 * selftest-diagnostic.c: New file.
9496 * selftest-diagnostic.h: New file.
9498 2017-11-02 James Bowman <james.bowman@ftdichip.com>
9500 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
9501 offset range for FT32B.
9502 * config/ft32/ft32.h: option "mcompress" enables relaxation.
9503 * config/ft32/ft32.md: Add TARGET_NOPM.
9504 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
9505 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
9507 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
9509 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
9511 2017-11-02 Jeff Law <law@redhat.com>
9513 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
9514 virtual keyword on FINAL OVERRIDE members.
9516 * tree-ssa-propagate.h (ssa_propagation_engine): Group
9517 virtuals together. Add virtual destructor.
9518 (substitute_and_fold_engine): Similarly.
9520 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
9522 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
9524 2017-11-02 Richard Biener <rguenther@suse.de>
9526 PR tree-optimization/82795
9527 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
9529 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9531 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
9533 (gcc_GAS_CHECK_FEATURE): Remove.
9534 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
9535 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
9536 * configure: Regenerate.
9538 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
9540 * config/arc/arc.c (hwloop_optimize): Account for empty
9543 2017-11-02 Richard Biener <rguenther@suse.de>
9546 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
9547 Truncate ARRAY_REF index and element size.
9549 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9551 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
9553 2017-11-01 Jeff Law <law@redhat.com>
9555 * tree-ssa-ccp.c (ccp_folder): New class derived from
9556 substitute_and_fold_engine.
9557 (ccp_folder::get_value): New member function.
9558 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
9559 (ccp_fold_stmt): Remove prototype.
9560 (ccp_finalize): Call substitute_and_fold from the ccp_class.
9561 * tree-ssa-copy.c (copy_folder): New class derived from
9562 substitute_and_fold_engine.
9563 (copy_folder::get_value): Renamed from get_value.
9564 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
9565 * tree-vrp.c (vrp_folder): New class derived from
9566 substitute_and_fold_engine.
9567 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
9568 (vrp_folder::get_value): New member function.
9569 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
9570 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
9571 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
9572 provide a class interface to folder/substitute routines.
9573 (ssa_prop_fold_stmt_fn): Remove typedef.
9574 (ssa_prop_get_value_fn): Likewise.
9575 (subsitute_and_fold): Remove prototype.
9576 (replace_uses_in): Likewise.
9577 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
9578 Renamed from replace_uses_in. Call the virtual member function
9579 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
9580 (substitute_and_fold_dom_walker): Remove initialization of
9581 data member entries for calbacks. Add substitute_and_fold_engine
9582 member and initialize it.
9583 (substitute_and_fold_dom_walker::before_dom_children0: Use the
9584 member functions for get_value, replace_phi_args_in c
9585 replace_uses_in, and fold_stmt calls.
9586 (substitute_and_fold_engine::substitute_and_fold): Renamed from
9587 substitute_and_fold. Remove assert. Update ctor call.
9589 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
9590 (ssa_prop_visit_phi_fn): Likewise.
9591 (class ssa_propagation_engine): New class to provide an interface
9593 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
9595 (ssa_prop_visit_phi): Likewise.
9596 (ssa_propagation_engine::simulate_stmt): Moved into class.
9597 Call visit_phi/visit_stmt from the class rather than via
9598 file scoped static variables.
9599 (ssa_propagation_engine::simulate_block): Moved into class.
9600 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
9601 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
9602 set file scoped statics for the visit_stmt/visit_phi callbacks.
9603 * tree-complex.c (complex_propagate): New class derived from
9604 ssa_propagation_engine.
9605 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
9606 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
9607 (tree_lower_complex): Call ssa_propagate via the complex_propagate
9609 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
9610 ssa_propagation_engine.
9611 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
9612 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
9613 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
9614 * tree-ssa-copy.c (copy_prop): New class derived from
9615 ssa_propagation_engine.
9616 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
9617 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
9618 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
9619 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
9620 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
9621 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
9622 (execute_vrp): Call ssa_propagate from the vrp_prop class.
9624 2017-11-01 Jakub Jelinek <jakub@redhat.com>
9626 PR rtl-optimization/82778
9627 PR rtl-optimization/82597
9628 * compare-elim.c (struct comparison): Add in_a_setter field.
9629 (find_comparison_dom_walker::before_dom_children): Remove killed
9630 bitmap and df_simulate_find_defs call, instead walk the defs.
9631 Compute last_setter and initialize in_a_setter. Merge definitions
9632 with first initialization for a few variables.
9633 (try_validate_parallel): Use insn_invalid_p instead of
9634 recog_memoized. Return insn rather than just the pattern.
9635 (try_merge_compare): Fix up comment. Don't uselessly test if
9636 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
9638 (execute_compare_elim_after_reload): Remove df_chain_add_problem
9641 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9642 Alan Hayward <alan.hayward@arm.com>
9643 David Sherwood <david.sherwood@arm.com>
9645 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
9646 aarch64_hard_regno_nregs to get the number of registers
9649 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9650 Alan Hayward <alan.hayward@arm.com>
9651 David Sherwood <david.sherwood@arm.com>
9653 * config/aarch64/constraints.md (Upl): Rename to...
9655 * config/aarch64/aarch64.md
9656 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
9659 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9660 Alan Hayward <alan.hayward@arm.com>
9661 David Sherwood <david.sherwood@arm.com>
9663 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
9664 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
9667 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9668 Alan Hayward <alan.hayward@arm.com>
9669 David Sherwood <david.sherwood@arm.com>
9671 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
9672 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
9673 (aarch64_evpc_dup): Generate rtl direcly, rather than using
9675 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
9676 of a single element.
9677 * config/aarch64/iterators.md: Add a comment above the permute
9678 unspecs to say that they are generated directly by
9679 aarch64_expand_vec_perm_const.
9680 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
9682 2017-11-01 Nathan Sidwell <nathan@acm.org>
9684 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
9686 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9688 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
9689 and medany code models, and describe what they do.
9691 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9693 Revert accidental duplicate:
9695 * combine.c (can_change_dest_mode): Reject changes in
9696 REGMODE_NATURAL_SIZE.
9698 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
9700 PR rtl-optimization/64682
9701 PR rtl-optimization/69567
9702 PR rtl-optimization/69737
9703 PR rtl-optimization/82683
9704 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
9705 register mentioned in the note, drop the note, unless it came from I3,
9706 in which case it should go to I3 again.
9708 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9710 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
9711 and return false if not.
9712 (clear_bytes_written_by, live_bytes_read): Update accordingly.
9714 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9716 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
9717 range is known to be empty.
9719 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9720 Alan Hayward <alan.hayward@arm.com>
9721 David Sherwood <david.sherwood@arm.com>
9723 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
9724 and CONST_VECTOR_NUNITS instead of computing the number of units from
9725 the byte sizes of the vector and element.
9726 (simplify_binary_operation_1): Likewise.
9727 (simplify_const_binary_operation): Likewise.
9728 (simplify_ternary_operation): Likewise.
9730 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9731 Alan Hayward <alan.hayward@arm.com>
9732 David Sherwood <david.sherwood@arm.com>
9734 * var-tracking.c (INT_MEM_OFFSET): Replace with...
9735 (int_mem_offset): ...this new function.
9736 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
9737 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
9738 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
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 * lower-subreg.c (interesting_mode_p): New function.
9746 (compute_costs, find_decomposable_subregs, decompose_register)
9747 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
9748 (resolve_clobber, dump_choices): Use it.
9750 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9751 Alan Hayward <alan.hayward@arm.com>
9752 David Sherwood <david.sherwood@arm.com>
9754 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
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 * alias.c (find_base_value, find_base_term): Only process integer
9761 truncations. Check the precision rather than the size.
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 * machmode.h (is_narrower_int_mode): New function
9768 * optabs.c (expand_float, expand_fix): Use it.
9769 * dwarf2out.c (rotate_loc_descriptor): Likewise.
9771 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9772 Alan Hayward <alan.hayward@arm.com>
9773 David Sherwood <david.sherwood@arm.com>
9775 * rtl.h (narrower_subreg_mode): New function.
9776 * ira-color.c (update_costs_from_allocno): Use it.
9778 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9779 Alan Hayward <alan.hayward@arm.com>
9780 David Sherwood <david.sherwood@arm.com>
9782 * optabs-query.h (convert_optab_p): New function, split out from...
9783 (convert_optab_handler): ...here.
9784 (widening_optab_handler): Delete.
9785 (find_widening_optab_handler): Remove permit_non_widening parameter.
9786 (find_widening_optab_handler_and_mode): Likewise. Provide an
9787 override that operates on mode class wrappers.
9788 * optabs-query.c (widening_optab_handler): Delete.
9789 (find_widening_optab_handler_and_mode): Remove permit_non_widening
9790 parameter. Assert that the two modes are the same class and that
9791 the "from" mode is narrower than the "to" mode. Use
9792 convert_optab_handler instead of widening_optab_handler.
9793 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
9794 instead of widening_optab_handler.
9795 * expr.c (expand_expr_real_2): Update calls to
9796 find_widening_optab_handler.
9797 * optabs.c (expand_widen_pattern_expr): Likewise.
9798 (expand_binop_directly): Take the insn_code as a parameter.
9799 (expand_binop): Only call find_widening_optab_handler for
9800 conversion optabs; use optab_handler otherwise. Update calls
9801 to find_widening_optab_handler and expand_binop_directly.
9802 Use convert_optab_handler instead of widening_optab_handler.
9803 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
9804 find_widening_optab_handler and use scalar_mode rather than
9806 (convert_plusminus_to_widen): Likewise.
9808 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9809 Alan Hayward <alan.hayward@arm.com>
9810 David Sherwood <david.sherwood@arm.com>
9812 * machmode.h (fixed_size_mode): New class.
9813 * rtl.h (get_pool_mode): Return fixed_size_mode.
9814 * gengtype.c (main): Add fixed_size_mode.
9815 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9816 (get_raw_arg_mode): Likewise.
9817 * doc/tm.texi: Regenerate.
9818 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9819 * targhooks.c (default_get_reg_raw_mode): Likewise.
9820 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9821 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9822 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9823 (msp430_get_raw_result_mode): Likewise.
9824 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9825 * dbxout.c (dbxout_parms): Require fixed-size modes.
9826 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9827 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9828 * omp-low.c (lower_oacc_reductions): Likewise.
9829 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9830 (simplify_subreg): Update accordingly.
9831 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9832 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9833 that aren't fixed-size.
9834 (get_pool_mode): Return a fixed_size_mode.
9835 (output_constant_pool_2): Take a fixed_size_mode.
9837 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9838 Alan Hayward <alan.hayward@arm.com>
9839 David Sherwood <david.sherwood@arm.com>
9841 * doc/rtl.texi (vec_series): Document.
9842 (const): Say that the operand can be a vec_series.
9843 * rtl.def (VEC_SERIES): New rtx code.
9844 * rtl.h (const_vec_series_p_1): Declare.
9845 (const_vec_series_p): New function.
9846 * emit-rtl.h (gen_const_vec_series): Declare.
9847 (gen_vec_series): Likewise.
9848 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9849 (gen_vec_series): Likewise.
9850 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9851 * simplify-rtx.c (simplify_unary_operation): Handle negations
9853 (simplify_binary_operation_series): New function.
9854 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9855 (test_vector_ops_series): New function.
9856 (test_vector_ops): Call it.
9857 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9858 gen_const_vec_series.
9859 (altivec_lvsr): Likewise.
9860 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9862 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9863 Alan Hayward <alan.hayward@arm.com>
9864 David Sherwood <david.sherwood@arm.com>
9866 * doc/rtl.texi (const): Update description of address constants.
9867 Say that vector constants are allowed too.
9868 * common.md (E, F): Use CONSTANT_P instead of checking for
9870 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9871 checking for CONST_VECTOR.
9872 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9874 * expr.c (expand_expr_real_2): Check for vector modes instead
9875 of checking for CONST_VECTOR.
9876 * rtl.h (const_vec_p): New function.
9877 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9878 (unwrap_const_vec_duplicate): Handle them here too.
9880 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9881 David Malcolm <dmalcolm@redhat.com>
9882 Alan Hayward <alan.hayward@arm.com>
9883 David Sherwood <david.sherwood@arm.com>
9885 * rtl.h (vec_duplicate_p): New function.
9886 * selftest-rtl.c (assert_rtx_eq_at): New function.
9887 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9888 (assert_rtx_eq_at): Declare.
9889 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9890 * selftest-run-tests.c (selftest::run_tests): Call it.
9891 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9892 (simplify_unary_operation_1): Recursively handle vector duplicates.
9893 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9895 (simplify_subreg): Handle subregs of vector duplicates.
9896 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9897 (selftest::simplify_rtx_c_tests): New functions.
9899 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9900 Alan Hayward <alan.hayward@arm.com>
9901 David Sherwood <david.sherwood@arm.com>
9903 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9904 (gen_vec_duplicate): Likewise.
9905 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9907 (gen_const_vector): ...here.
9908 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9909 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9910 whose elements are all equal.
9911 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
9912 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
9913 (simplify_relational_operation): Likewise.
9914 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
9916 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
9917 (aarch64_expand_vector_init): Likewise.
9918 * config/arm/arm.c (neon_vdup_constant): Likewise.
9919 (neon_expand_vector_init): Likewise.
9920 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
9921 (arm_block_set_unaligned_vect): Likewise.
9922 (arm_block_set_aligned_vect): Likewise.
9923 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
9924 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
9925 (expand_vec_perm_even_odd_pack): Likewise.
9926 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
9927 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
9928 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
9929 gen_const_vec_duplicate.
9930 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
9931 * config/mips/mips.c (mips_gen_const_int_vector): Use
9932 gen_const_vec_duplicate.
9933 (mips_expand_vector_init): Use CONST0_RTX.
9934 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
9935 (define_split): Use gen_const_vec_duplicate.
9936 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
9937 (define_split): Use gen_const_vec_duplicate.
9938 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
9939 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
9940 * config/spu/spu.c (spu_const): Likewise.
9942 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9943 Alan Hayward <alan.hayward@arm.com>
9944 David Sherwood <david.sherwood@arm.com>
9946 * combine.c (can_change_dest_mode): Reject changes in
9947 REGMODE_NATURAL_SIZE.
9949 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
9951 * configure.ac (--enable-libssp): New.
9952 (gcc_cv_libc_provides_ssp): Check for explicit setting before
9953 trying to determine target-specific default. Adjust indentation.
9954 * configure: Regenerated.
9955 * doc/install.texi (Configuration): Expand --disable-libssp
9958 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
9960 config/i386/i386.c (ix86_expand_epilogue): Correct stack
9963 2017-10-31 Martin Jambor <mjambor@suse.cz>
9966 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
9968 2017-10-31 David Malcolm <dmalcolm@redhat.com>
9970 * auto-profile.c (autofdo_source_profile::read): Use
9971 UNKNOWN_LOCATION rather than 0.
9972 * diagnostic-core.h (warning_at_rich_loc): Rename to...
9973 (warning_at): ...this overload.
9974 (warning_at_rich_loc_n): Rename to...
9975 (warning_n): ...this overload.
9976 (error_at_rich_loc): Rename to...
9977 (error_at): ...this overload.
9978 (pedwarn_at_rich_loc): Rename to...
9979 (pedwarn): ...this overload.
9980 (permerror_at_rich_loc): Rename to...
9981 (permerror): ...this overload.
9982 (inform_at_rich_loc): Rename to...
9983 (inform): ...this overload.
9984 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
9985 (diagnostic_n_impl_richloc): Rename to...
9986 (diagnostic_n_impl): ...this rich_location *-based decl.
9987 (inform_at_rich_loc): Rename to...
9988 (inform): ...this, and add an assertion.
9989 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
9990 (warning_at_rich_loc): Rename to...
9991 (warning_at): ...this, and add an assertion.
9992 (warning_at_rich_loc_n): Rename to...
9993 (warning_n): ...this, and add an assertion.
9994 (warning_n): Update location_t-based implementation for removal of
9995 location_t-based diagnostic_n_impl.
9996 (pedwarn_at_rich_loc): Rename to...
9997 (pedwarn): ...this, and add an assertion.
9998 (permerror_at_rich_loc): Rename to...
9999 (permerror): ...this, and add an assertion.
10000 (error_n): Update for removal of location_t-based diagnostic_n_impl.
10001 (error_at_rich_loc): Rename to...
10002 (error_at): ...this, and add an assertion.
10003 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
10004 (driver::do_spec_on_infiles): Likewise.
10005 * substring-locations.c (format_warning_va): Update for renaming
10006 of inform_at_rich_loc.
10008 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
10010 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
10011 _Float<N>X built-in functions so that the variant without the
10012 "__builtin_" prefix is only enabled for the GNU C and Objective C
10013 languages when they are in non-strict ANSI/ISO mode.
10014 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
10015 * target.def (floatn_builtin_p): Add a target hook to control
10016 whether _Float<N> and _Float<N>X built-in functions without the
10017 "__builtin_" prefix are enabled, and return true for C and
10018 Objective C in the default hook. Include langhooks.h in
10020 * targhooks.h (default_floatn_builtin_p): Likewise.
10021 * targhooks.c (default_floatn_builtin_p): Likewise.
10022 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
10023 floatn_builtin_p target hook.
10024 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
10026 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
10027 Eric Botcazou <ebotcazou@adacore.com>
10029 PR rtl-optimization/81803
10030 * lra-constraints.c (curr_insn_transform): Also reload the whole
10031 register for a strict subreg no wider than a word if this is for
10032 a WORD_REGISTER_OPERATIONS target.
10034 2017-10-31 Jason Merrill <jason@redhat.com>
10036 * gdbinit.in: Skip over inlines from timevar.h.
10038 2017-10-31 Martin Liska <mliska@suse.cz>
10040 * doc/gcov.texi: Document new option.
10041 * gcov.c (print_usage): Likewise print it.
10042 (process_args): Support the argument.
10043 (format_count): New function.
10044 (format_gcov): Use the function.
10046 2017-10-31 Martin Liska <mliska@suse.cz>
10048 * gcov.c (struct name_map): do not use typedef.
10049 Define operator== and operator<.
10050 (name_search): Remove.
10051 (name_sort): Remove.
10052 (main): Do not allocate names.
10053 (process_file): Add vertical space.
10054 (generate_results): Use std::find.
10055 (release_structures): Do not release memory.
10056 (find_source): Use std::find.
10058 2017-10-31 Martin Liska <mliska@suse.cz>
10060 * gcov.c (struct line_info): Remove it's typedef.
10061 (line_info::line_info): Add proper ctor.
10062 (line_info::has_block): Do not use a typedef.
10063 (struct source_info): Do not use typedef.
10064 (circuit): Likewise.
10065 (get_cycles_count): Likewise.
10066 (output_intermediate_file): Iterate via vector iterator.
10067 (add_line_counts): Use std::vector methods.
10068 (accumulate_line_counts): Likewise.
10069 (output_lines): Likewise.
10071 2017-10-31 Martin Liska <mliska@suse.cz>
10073 * gcov.c (struct source_info): Remove typedef.
10074 (source_info::source_info): Add proper ctor.
10075 (accumulate_line_counts): Use struct, not it's typedef.
10076 (output_gcov_file): Likewise.
10077 (output_lines): Likewise.
10078 (main): Do not allocate an array.
10079 (output_intermediate_file): Use size of vector container.
10080 (process_file): Resize the vector.
10081 (generate_results): Do not preallocate, use newly added vector
10083 (release_structures): Do not release sources.
10084 (find_source): Use vector methods.
10085 (add_line_counts): Do not use typedef.
10087 2017-10-31 Martin Liska <mliska@suse.cz>
10089 * doc/gcov.texi: Document that.
10090 * gcov.c (add_line_counts): Mark lines with a non-executed
10092 (output_line_beginning): Handle such lines.
10093 (output_lines): Pass new argument.
10094 (output_intermediate_file): Print it in intermediate format.
10096 2017-10-31 Martin Liska <mliska@suse.cz>
10098 * color-macros.h: New file.
10099 * diagnostic-color.c: Factor out color related to macros to
10101 * doc/gcov.texi: Document -k option.
10102 * gcov.c (INCLUDE_STRING): Include string.h.
10103 (print_usage): Add -k option.
10104 (process_args): Parse it.
10105 (pad_count_string): New function.
10106 (output_line_beginning): Likewise.
10107 (DEFAULT_LINE_START): New macro.
10108 (output_lines): Support color output.
10110 2017-10-31 Martin Liska <mliska@suse.cz>
10112 PR gcov-profile/82633
10113 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
10114 their interaction with GCOV infrastructure.
10115 * configure.ac: Add -fkeep-{inline,static}-functions to
10117 * configure: Regenerate.
10119 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
10122 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
10124 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
10127 * config/rs6000/rs6000.md (allocate_stack): Force update interval
10128 into a register if it does not fit into an immediate offset field.
10130 2017-10-31 Olivier Hainque <hainque@adacore.com>
10132 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
10134 2017-10-31 Julia Koval <julia.koval@intel.com>
10136 * config.gcc: Add gfniintrin.h.
10137 * config/i386/gfniintrin.h: New.
10138 * config/i386/i386-builtin-types.def
10139 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
10140 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
10141 __builtin_ia32_vgf2p8affineinvqb_v32qi,
10142 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
10143 __builtin_ia32_vgf2p8affineinvqb_v16qi,
10144 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
10145 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
10146 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
10147 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
10148 V64QI_FTYPE_V64QI_V64QI_INT): New types.
10149 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
10150 * config/i386/immintrin.h: Include gfniintrin.h.
10151 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
10153 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
10155 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
10157 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10159 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
10160 (arm_ashldi3_1bit): Remove pattern.
10161 (ashrdi3): Remove shift by 1 expansion.
10162 (arm_ashrdi3_1bit): Remove pattern.
10163 (lshrdi3): Remove shift by 1 expansion.
10164 (arm_lshrdi3_1bit): Remove pattern.
10165 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
10166 cost of ashldi3 by 1.
10167 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
10168 (<shift>di3_neon): Likewise.
10170 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
10172 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
10173 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
10174 and (*aarch64_simd_mov<VQ:mode>).
10175 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
10176 pattern alternative.
10177 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
10178 attributes to match pattern alternative.
10180 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
10182 * config.gcc (powerpc*-*-*): Add emmintrin.h.
10183 * config/rs6000/emmintrin.h: New file.
10184 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
10186 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10188 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
10189 * (movdi_vfp_cortexa8): Remove pattern.
10191 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10193 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
10195 (Specific, alpha*-dec-osf5.1): Remove.
10196 (Specific, mips-sgi-irix5): Remove.
10197 (Specific, mips-sgi-irix6): Remove.
10199 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10201 PR middle-end/22141
10202 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
10203 arguments to clear_bit_region_be.
10205 2017-10-30 Jim Wilson <wilson@tuliptree.org>
10207 * gimplify.c: Include memmodel.h.
10209 2017-10-30 Martin Jambor <mjambor@suse.cz>
10211 * omp-grid.c (grid_attempt_target_gridification): Also insert a
10212 condition whether loop should be executed at all.
10214 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
10216 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
10217 gimple folding of vec_madd() intrinsics.
10218 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
10219 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
10220 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
10222 2017-10-30 Richard Biener <rguenther@suse.de>
10224 PR tree-optimization/82762
10226 2017-10-23 Richard Biener <rguenther@suse.de>
10228 PR tree-optimization/82129
10230 2017-08-01 Richard Biener <rguenther@suse.de>
10232 PR tree-optimization/81181
10233 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10234 (compute_antic): ... end of iteration here.
10236 2017-10-30 Joseph Myers <joseph@codesourcery.com>
10238 * doc/invoke.texi (C Dialect Options): Document -std=c17,
10239 -std=iso9899:2017 and -std=gnu17.
10240 * doc/standards.texi (C Language): Document C17 support.
10241 * doc/cpp.texi (Overview): Mention -std=c17.
10242 (Standard Predefined Macros): Document C11 and C17 values of
10243 __STDC_VERSION__. Do not refer to C99 support as incomplete.
10244 * doc/extend.texi (Inline): Do not list individual options for
10245 standards newer than C99.
10246 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
10248 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
10251 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
10253 * asan.c (asan_finish_file): Align asan globals array by shadow
10256 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10258 PR middle-end/22141
10259 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
10260 (struct store_immediate_info): Add bitregion_start and bitregion_end
10262 (store_immediate_info::store_immediate_info): Add brs and bre
10263 arguments and initialize bitregion_{start,end} from those.
10264 (struct merged_store_group): Add bitregion_start, bitregion_end,
10265 align_base and mask fields. Drop unnecessary struct keyword from
10266 struct store_immediate_info. Add do_merge method.
10267 (clear_bit_region_be): Use memset instead of loop storing zeros.
10268 (merged_store_group::do_merge): New method.
10269 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
10270 stores as long as the surrounding bitregions have no gaps.
10271 (merged_store_group::merge_overlapping): Use do_merge.
10272 (merged_store_group::apply_stores): Test that bitregion_{start,end}
10273 is byte aligned, rather than requiring that start and width are
10274 byte aligned. Drop unnecessary struct keyword from
10275 struct store_immediate_info. Allocate and populate also mask array.
10276 Make start of the arrays relative to bitregion_start rather than
10277 start and size them according to bitregion_{end,start} difference.
10278 (struct imm_store_chain_info): Drop unnecessary struct keyword from
10279 struct store_immediate_info.
10280 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
10281 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
10282 struct keyword from struct store_immediate_info.
10283 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
10284 between stores as long as the surrounding bitregions have no gaps.
10286 (struct split_store): Add orig non-static data member.
10287 (split_store::split_store): Initialize orig to false.
10288 (find_constituent_stmts): Return store_immediate_info *, non-NULL
10289 if there is exactly a single original stmt. Change stmts argument
10290 to pointer from reference, if NULL, don't push anything to it. Add
10291 first argument, use it to optimize skipping over orig stmts that
10292 are known to be before bitpos already. Simplify.
10293 (split_group): Return unsigned int count how many stores are or
10294 would be needed rather than a bool. Add allow_unaligned argument.
10295 Change split_stores argument from reference to pointer, if NULL,
10296 only do a dry run computing how many stores would be produced.
10297 Rewritten algorithm to use both alignment and misalign if
10298 !allow_unaligned and handle bitfield stores with gaps.
10299 (imm_store_chain_info::output_merged_store): Set start_byte_pos
10300 from bitregion_start instead of start. Compute allow_unaligned
10301 here, if true, do 2 split_group dry runs to compute which one
10302 produces fewer stores and prefer aligned if equal. Punt if
10303 new count is bigger or equal than original before emitting any
10304 statements, rather than during that. Remove no longer needed
10305 new_ssa_names tracking. Replace num_stmts with
10306 split_stores.length (). Use 32-bit stack allocated entries
10307 in split_stores auto_vec. Try to reuse original store lhs/rhs1
10308 if possible. Handle bitfields with gaps.
10309 (pass_store_merging::execute): Ignore bitsize == 0 stores.
10310 Compute bitregion_{start,end} for the stores and construct
10311 store_immediate_info with that. Formatting fixes.
10313 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
10316 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
10317 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
10319 2017-10-29 Jim Wilson <wilson@tuliptree.org>
10321 * gimplify.c: Include tm_p.h.
10323 * common.opt (gcoff): Re-add as ignored option.
10324 (gcoff1, gcoff2, gcoff3): Likewise.
10326 * Makefile.in (OBJS): Delete sdbout.o.
10327 (GTFILES): Delete $(srcdir)/sdbout.c.
10328 * debug.h: Delete sdb_debug_hooks.
10329 * final.c: Delete sdbout.h include.
10330 (final_scan_insn): Delete SDB_DEBUG check.
10331 (rest_of_clean_state): Likewise.
10332 * output.h: Delete sdb_begin_function_line.
10333 * sdbout.c: Delete.
10334 * sdbout.h: Delete.
10335 * toplev.c: Delete sdbout.h include.
10336 (process_options): Delete SDB_DEBUG check.
10337 * tree-core.h (tree_type_common): Delete pointer field of
10339 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
10340 TYPE_SYMTAB_POINTER.
10341 * tree.h (TYPE_SYMTAB_POINTER): Delete.
10342 (TYPE_SYMTAB_IS_POINTER): Delete.
10343 (TYPE_SYMTAB_IS_DIE): Renumber.
10344 * xcoffout.c: Refer to former sdbout.c file.
10345 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
10347 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
10348 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
10349 (Debugging Options): Delete -gcoff.
10350 (-gstabs): Delete SDB reference.
10352 (-gcoff@var{level}): Delete.
10353 * doc/passes.texi (Debugging information output): Delete SDB and
10354 sdbout.c references.
10355 * doc/tm.texi: Regenerate.
10356 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
10357 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
10359 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
10360 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
10361 (SDB_DEBUGGING_INFO): Delete.
10362 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
10363 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
10364 * target.def (output_source_filename): Delete COFF reference.
10366 * common.opt (gcoff): Delete.
10367 (gxcoff+): Update Negative chain.
10368 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
10370 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
10371 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
10372 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
10373 and SDB references.
10374 (expand_function_start): Change sdb reference to past tense.
10375 (expand_function_end): Change sdb reference to past tense.
10376 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
10377 * opts.c (debug_type_names): Delete coff entry.
10378 (common_handle_option): Delete OPT_gcoff case.
10379 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
10381 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
10382 * config/cris/cris.h: Delete SDB reference in comment.
10383 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
10384 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
10385 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
10386 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
10388 (ix86_expand_prologue): Likewise.
10389 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
10390 * config/ia64/ia64.h: Likewise.
10391 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
10392 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
10394 * config/mmix/mmix.h: Likewise.
10395 * config/nds32/nds32.c: Likewise.
10396 * config/stormy/storym16.h: Likewise.
10397 * config/visium/visium.h: Likewise.
10398 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
10400 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
10402 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
10403 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
10404 FRAME_POINTER_REGNUM point at high end of local var area.
10406 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
10408 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
10409 Move comment around. Do not reset best_edge for a copiable
10410 destination if the copy would cause a partition change.
10411 (better_edge_p): Remove redundant check.
10413 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10415 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
10417 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
10419 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
10420 for math functions that have _Float<N> and _Float<N>X variants.
10421 (mathfn_built_in_2): Add support for math functions that have
10422 _Float<N> and _Float<N>X variants.
10423 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
10424 (expand_builtin_mathfn_ternary): Add support for fma with
10425 _Float<N> and _Float<N>X variants.
10426 (expand_builtin): Likewise.
10427 (fold_builtin_3): Likewise.
10428 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
10429 create math function _Float<N> and _Float<N>X variants as external
10431 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
10432 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
10433 the __builtin_ prefix and if not strict ansi, without the prefix.
10434 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
10435 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
10436 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
10437 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
10438 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
10439 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
10440 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
10441 function signatures for fma _Float<N> and _Float<N>X variants.
10442 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
10443 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
10444 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
10445 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
10446 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
10447 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
10448 * gencfn-macros.c (print_case_cfn): Add support for math functions
10449 that have _Float<N> and _Float<N>X variants.
10450 (print_define_operator_list): Likewise.
10451 (fltfn_suffixes): Likewise.
10453 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
10454 for math functions that have _Float<N> and _Float<N>X variants.
10455 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
10456 and _Float<N>X variants.
10457 (COPYSIGN): Likewise.
10460 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
10461 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
10463 (integer_valued_read_call_p): Likewise.
10464 * fold-const-call.c (fold_const_call_ss): Likewise.
10465 (fold_const_call_sss): Add support for copysign, fmin, and fmax
10466 _Float<N> and _Float<N>X variants.
10467 (fold_const_call_ssss): Add support for fma _Float<N> and
10468 _Float<N>X variants.
10469 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
10470 support for copysign and fma _Float<N> and _Float<N>X variants.
10471 (backprop::process_builtin_call_use): Likewise.
10472 * tree-call-cdce.c (can_test_argument_range); Add support for
10473 sqrt _Float<N> and _Float<N>X variants.
10474 (edom_only_function): Likewise.
10475 (get_no_error_domain): Likewise.
10476 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
10477 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
10478 copysign _Float<N> and _Float<N>X variants.
10479 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
10480 handled by machine independent code.
10481 (FMAF128): Likewise.
10482 * doc/cpp.texi (Common Predefined Macros): Document defining
10483 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
10484 fma _Float<N> and _Float<N>X variants.
10486 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10489 * config/i386/i386-modes.def (CCFPU): Remove definition.
10490 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
10491 (ix86_cc_modes_compatible): Ditto.
10492 (ix86_expand_carry_flag_compare): Ditto.
10493 (ix86_expand_int_movcc): Ditto.
10494 (ix86_expand_int_addcc): Ditto.
10495 (ix86_reverse_condition): Ditto.
10496 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
10497 Return true/false for unordered/ordered fp comparisons.
10498 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
10499 (ix86_prepare_fp_compare_args): Update for rename.
10500 (ix86_expand_fp_compare): Update for rename. Generate unordered
10501 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
10502 (ix86_expand_sse_compare_and_jump): Ditto.
10503 * config/i386/predicates.md (fcmov_comparison_operator):
10504 Remove CCFPU mode handling.
10505 (ix86_comparison_operator): Ditto.
10506 (ix86_carry_flag_operator): Ditto.
10507 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
10508 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
10509 (*cmpu<mode>_cc_i387): Ditto.
10510 (FPCMP): Remove mode iterator.
10511 (unord): Remove mode attribute.
10512 (unord_subst): New define_subst transformation
10513 (unord): New define_subst attribute.
10514 (unordered): Ditto.
10515 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
10516 (*cmpi<unord>xf_i387): Ditto.
10517 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
10518 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
10519 using unord_subst transformation.
10520 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
10521 (round_saeonly): Also handle CCFP mode.
10522 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
10523 Remove UNSPEC_SAHF unspec handling.
10525 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10527 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
10529 2017-10-27 Jeff Law <law@redhat.com>
10531 * gimple-ssa-sprintf.c: Include domwalk.h.
10532 (class sprintf_dom_walker): New class, derived from dom_walker.
10533 (sprintf_dom_walker::before_dom_children): New function.
10534 (struct call_info): Moved into sprintf_dom_walker class
10535 (compute_formath_length, handle_gimple_call): Likewise.
10536 (sprintf_length::execute): Call the dominator walker rather
10537 than walking the statements.
10539 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
10540 gimple statement locations.
10541 (check_array_bounds): Corresponding changes. Get the statement's
10542 location directly from wi->stmt.
10544 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
10547 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
10549 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10551 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
10552 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
10554 2017-10-27 Jakub Jelinek <jakub@redhat.com>
10557 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
10558 * config/i386/i386.c (maybe_get_pool_constant): Removed.
10559 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
10560 maybe_get_pool_constant.
10561 * config/i386/predicates.md (zero_extended_scalar_load_operand):
10564 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10566 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
10567 2.26 caveat. Update gas and gld versions.
10568 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
10571 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10573 * cgraph.h (set_malloc_flag): Declare.
10574 * cgraph.c (set_malloc_flag_1): New function.
10575 (set_malloc_flag): Likewise.
10576 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
10577 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
10579 (read_ipa_call_summary): Add support for reading is_return_callee.
10580 (write_ipa_call_summary): Stream is_return_callee.
10581 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
10582 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
10583 ipa-prop.h, ipa-fnsummary.h.
10584 (pure_const_names): Change to static.
10585 (malloc_state_e): Define.
10586 (malloc_state_names): Define.
10587 (funct_state_d): Add field malloc_state.
10588 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
10589 (check_retval_uses): New function.
10590 (malloc_candidate_p): Likewise.
10591 (analyze_function): Add support for malloc attribute.
10592 (pure_const_write_summary): Stream malloc_state.
10593 (pure_const_read_summary): Add support for reading malloc_state.
10594 (dump_malloc_lattice): New function.
10595 (propagate_malloc): New function.
10596 (warn_function_malloc): New function.
10597 (ipa_pure_const::execute): Call propagate_malloc and
10598 ipa_free_fn_summary.
10599 (pass_local_pure_const::execute): Add support for malloc attribute.
10600 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
10601 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
10603 2017-10-27 Martin Liska <mliska@suse.cz>
10605 PR gcov-profile/82457
10606 * doc/invoke.texi: Document that one needs a non-strict ISO mode
10607 for fork-like functions to be properly instrumented.
10609 2017-10-27 Richard Biener <rguenther@suse.de>
10611 PR middle-end/81659
10612 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
10613 info when we redirected EH.
10615 2017-10-26 Michael Collison <michael.collison@arm.com>
10617 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
10619 (<optab>_trunchf<GPI:mode>2: New pattern.
10620 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
10621 * config/aarch64/iterators.md (wv): New mode attribute.
10622 (vf, VF): New mode attributes.
10623 (vgp, VGP): New mode attributes.
10624 (s): Update attribute with SImode and DImode prefixes.
10626 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10628 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
10629 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
10630 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
10631 (nios2_option_overide): Initialize it. Don't allow R0-relative
10632 addressing with PIC.
10633 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
10634 (nios2_symbolic_constant_p): Likewise.
10635 (nios2_legitimate_address_p): Likewise.
10636 (nios2_r0rel_section_name_p): New.
10637 (nios2_symbol_ref_in_r0rel_data_p): New.
10638 (nios2_emit_move_sequence): Handle r0rel_constant_p.
10639 (r0rel_constant_p): New.
10640 (nios2_print_operand_address): Handle r0rel_constant_p.
10641 (nios2_cdx_narrow_form_p): Likewise.
10642 * config/nios2/nios2.opt (mr0rel-sec=): New option.
10643 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
10644 (Nios II Options): Document -mr0rel-sec.
10646 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10648 * config/nios2/nios2.c: Include xregex.h.
10649 (nios2_gprel_sec_regex): New.
10650 (nios2_option_overide): Initialize it. Don't allow GP-relative
10651 addressing with PIC.
10652 (nios2_small_section_name_p): Check for regex match.
10653 * config/nios2/nios2.opt (mgprel-sec=): New option.
10654 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
10655 (Nios II Options): Document -mgprel-sec.
10657 2017-10-26 Jim Wilson <wilson@tuliptree.org>
10659 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
10661 2017-10-26 Tom de Vries <tom@codesourcery.com>
10663 PR tree-optimization/82707
10664 * gimple.c (gimple_copy): Fix unsharing of
10665 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
10667 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
10669 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
10670 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
10671 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
10672 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
10673 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
10674 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
10675 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
10676 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
10677 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
10678 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
10679 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
10680 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
10681 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
10682 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
10683 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
10684 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
10685 _mm512_mask_cmpunord_ps_mask): New intrinsics.
10687 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10689 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
10691 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
10692 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
10693 warning to rs6000.c. Remove the Undocumented flag, since it has
10695 (-mabi=ibmlongdouble): Likewise.
10696 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
10697 already set, set the default format for long double.
10698 (rs6000_debug_reg_global): Print whether long double is IBM or
10700 (rs6000_option_override_internal): Rework setting long double
10701 format. Only warn if the user is changing the long double default
10702 and they did not use -Wno-psabi.
10703 * doc/invoke.texi (PowerPC options): Update the documentation for
10704 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
10706 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10707 Alan Hayward <alan.hayward@arm.com>
10708 David Sherwood <david.sherwood@arm.com>
10710 * rtl.h (wider_subreg_mode): New function.
10711 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
10712 rather than an unsigned int *.
10713 * ira-color.c (regno_max_ref_width): Replace with...
10714 (regno_max_ref_mode): ...this new variable.
10715 (coalesced_pseudo_reg_slot_compare): Update accordingly.
10716 Use wider_subreg_mode.
10717 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
10718 rather than an unsigned int *.
10719 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
10720 (process_alt_operands): Likewise.
10721 (invariant_p): Likewise.
10722 * lra-spills.c (assign_mem_slot): Likewise.
10723 (add_pseudo_to_slot): Likewise.
10724 * lra.c (collect_non_operand_hard_regs): Likewise.
10725 (add_regs_to_insn_regno_info): Likewise.
10726 * reload1.c (regno_max_ref_width): Replace with...
10727 (regno_max_ref_mode): ...this new variable.
10728 (reload): Update accordingly. Update call to
10729 ira_sort_regnos_for_alter_reg.
10730 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
10731 (init_eliminable_invariants): Update to use regno_max_ref_mode.
10732 (scan_paradoxical_subregs): Likewise.
10734 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10736 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
10737 (aarch64_frame): Add emit_frame_chain boolean.
10738 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10739 Move eh_return case to aarch64_layout_frame.
10740 (aarch64_layout_frame): Initialize emit_frame_chain.
10741 (aarch64_expand_prologue): Use emit_frame_chain.
10743 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10745 * config/aarch64/aarch64.c (aarch64_layout_frame):
10746 Ensure LR is always stored at the bottom of the callee-saves.
10747 Remove rarely used frame layout which saves callee-saves at top of
10748 frame, so the store of LR can be used as a valid probe in all cases.
10750 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10752 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
10753 Improve unaligned TImode/TFmode base/offset split.
10755 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10756 Alan Hayward <alan.hayward@arm.com>
10757 David Sherwood <david.sherwood@arm.com>
10759 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
10760 * combine.c (find_single_use_1): Likewise.
10761 (expand_field_assignment): Likewise.
10762 (move_deaths): Likewise.
10763 * lra-constraints.c (simplify_operand_subreg): Likewise.
10764 (curr_insn_transform): Likewise.
10765 * lra.c (collect_non_operand_hard_regs): Likewise.
10766 (add_regs_to_insn_regno_info): Likewise.
10767 * rtlanal.c (reg_referenced_p): Likewise.
10768 (covers_regno_no_parallel_p): Likewise.
10770 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10772 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
10773 Don't print any bits outside the precision of the value.
10774 * wide-int.cc (test_printing): Add some new tests.
10776 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10778 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
10779 supports -xbrace_comment option.
10780 * configure: Regenerate.
10781 * config.in: Regenerate.
10782 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
10783 (ASM_CPU_SPEC): Use it.
10785 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10787 * target.def (static_rtx_alignment): New hook.
10788 * targhooks.h (default_static_rtx_alignment): Declare.
10789 * targhooks.c (default_static_rtx_alignment): New function.
10790 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
10791 * doc/tm.texi: Regenerate.
10792 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
10793 instead of targetm.constant_alignment. Remove call to
10794 set_mem_attributes.
10795 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10796 (cris_preferred_mininum_alignment): New function, split out from...
10797 (cris_constant_alignment): ...here.
10798 (cris_static_rtx_alignment): New function.
10799 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
10801 (ix86_constant_alignment): ...here.
10802 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10803 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10804 (mmix_static_rtx_alignment): New function.
10805 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10806 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10808 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10811 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10812 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10814 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10816 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10817 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10818 mark as a sign-extending load.
10819 (local_pic_load_u): Define.
10821 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10823 PR middle-end/82062
10824 * fold-const.c (operand_equal_for_comparison_p): Also return true
10825 if ARG0 is a simple variant of ARG1 with narrower precision.
10826 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10828 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10830 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10832 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10833 scatter_static, scatter_per_elt.
10834 * x86-tune-costs.h: Add new cost entries.
10836 2017-10-25 Richard Biener <rguenther@suse.de>
10838 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10839 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10840 class pass_fre): Move to ...
10841 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10842 class pass_fre): ... here and adjust for statistics.
10844 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10847 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10848 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10849 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10850 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10853 2017-10-25 Richard Biener <rguenther@suse.de>
10855 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10856 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10857 eliminate_push_avail, eliminate_insert): Move inside...
10858 (class eliminate_dom_walker): ... this class in preparation
10860 (fini_eliminate): Remove by merging with ...
10861 (eliminate): ... this function. Adjust for class changes.
10862 (pass_pre::execute): Remove fini_eliminate call.
10863 (pass_fre::execute): Likewise.
10865 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10868 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10869 (VPERMI2, VPERMI2I): New mode iterators.
10870 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10871 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10873 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10874 mode iterator. Remove 3 old define_insn patterns.
10875 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10876 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10877 VPERMI2 mode iterator, remove the other two expanders.
10878 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10879 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10880 instructions, remove the other two patterns.
10881 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10882 mode iterator, remove the other two patterns.
10883 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10884 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10885 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10886 and adjust argument order accordingly.
10887 (ix86_expand_vec_perm): Adjust caller.
10888 (expand_vec_perm_1): Likewise.
10889 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10890 (expand_vec_perm_vpermt2_vpshub2): ... this.
10891 (ix86_expand_vec_perm_const_1): Adjust caller.
10892 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10895 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10896 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10897 (vec_shl_<mode>): Remove unused expander.
10898 (avx512bw_<shift_insn><mode>3): New define_insn.
10899 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10900 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10902 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10905 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10908 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10910 PR rtl-optimization/82396
10911 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
10912 (autopref_multipass_init): Simplify initialization.
10913 (autopref_rank_data): Simplify sort order.
10914 * gcc/sched-int.h (autopref_multipass_data_): Remove
10915 multi_mem_insn_p, min_offset and max_offset.
10917 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10919 PR middle-end/60580
10920 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10921 Check special value of flag_omit_frame_pointer.
10922 (aarch64_can_eliminate): Likewise.
10923 (aarch64_override_options_after_change_1): Simplify handling of
10924 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
10926 2017-10-24 Richard Biener <rguenther@suse.de>
10928 PR tree-optimization/82697
10929 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
10930 zero for conditional load and unconditional store.
10932 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10934 * doc/install.texi: Document bootstrap-cet.
10936 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10939 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
10940 ENDBR instruction at function entrance if function is only
10943 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10946 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
10947 patterns to better describe from which operation the CF is computed.
10948 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
10949 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
10950 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
10951 is 0, use _0 suffixed expanders instead of emitting a comparison
10954 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
10956 * config/i386/i386.md(*movsf_internal, *movdf_internal):
10957 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
10959 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
10961 PR middle-end/82569
10962 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
10963 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
10964 * loop-iv.c (iv_get_reaching_def): Likewise.
10965 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
10966 variable is promoted and the partition contains undefined values.
10968 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10970 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
10972 (nios2_address_cost): Define.
10973 (nios2_legitimize_address): Recognize (exp + constant) directly.
10974 (TARGET_ADDRESS_COST): Define.
10976 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10978 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
10979 (nios2_symbolic_memory_operand_p): Declare.
10980 (nios2_split_large_constant): Declare.
10981 (nios2_split_symbolic_memory_operand): Declare.
10982 * config/nios2/nios2.c: Adjust includes.
10983 (nios2_symbolic_constant_allowed): New.
10984 (nios2_symbolic_constant_p): New.
10985 (nios2_plus_symbolic_constant_p): New.
10986 (nios2_valid_addr_expr_p): Recognize addresses involving
10987 symbolic constants.
10988 (nios2_legitimate_address_p): Likewise, also LO_SUM.
10989 (nios2_symbolic_memory_operand_p): New.
10990 (nios2_large_constant_p): New.
10991 (nios2_split_large_constant): New.
10992 (nios2_split_plus_large_constant): New.
10993 (nios2_split_symbolic_memory_operand): New.
10994 (nios2_legitimize_address): Code refactoring. Handle addresses
10995 involving symbolic constants.
10996 (nios2_emit_move_sequence): Likewise.
10997 (nios2_print_operand): Improve error output.
10998 (nios2_print_operand_address): Handle LO_SUM.
10999 (nios2_cdx_narrow_form_p): Likewise.
11000 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
11001 operands involving symbolic constants.
11002 (movhi_internal, movsi_internal): Likewise.
11003 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
11004 (extendhisi2, extendqi<mode>2): Likewise.
11006 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11008 * tree-pass.h (PROP_rtl_split_insns): Define.
11009 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
11011 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11013 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
11015 2017-10-23 Jakub Jelinek <jakub@redhat.com>
11018 * target.def (const_not_ok_for_debug_p): Default to
11019 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
11020 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
11021 * targhooks.c (default_const_not_ok_for_debug_p): New function.
11022 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
11023 which targetm.const_not_ok_for_debug_p returned true.
11024 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
11026 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
11028 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
11029 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
11030 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
11032 (ix86_const_not_ok_for_debug_p): New function.
11033 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
11034 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
11036 2017-10-23 David Malcolm <dmalcolm@redhat.com>
11039 * system.h: Conditionally include "unique-ptr.h" if
11040 INCLUDE_UNIQUE_PTR is defined.
11041 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
11042 of defining INCLUDE_UNIQUE_PTR before including "system.h".
11044 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
11046 * config/rl78/rl78.md: New define_expand "subdi3".
11048 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
11051 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
11052 DF_REF_INSN if DF_REF_INSN_INFO is false.
11054 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
11056 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
11057 xmm_move instead of sse_move.
11058 (sse_store_index): New function.
11059 (ix86_register_move_cost): Be more sensible about mismatch stall;
11060 model AVX moves correctly; make difference between sse->integer and
11062 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
11063 moves; make difference between SSE and AVX.
11064 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
11065 and zmm_move. Increase size of sse load and store tables;
11066 add unaligned load and store tables; add ssemmx_to_integer.
11067 * x86-tune-costs.h: Update all entries according to real
11068 move latencies from Agner Fog's manual and chip documentation.
11070 2017-10-23 Jakub Jelinek <jakub@redhat.com>
11073 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
11074 * config/i386/constraints.md (Wf): New constraint.
11075 * config/i386/i386.md (UNSPEC_SBB): New unspec.
11076 (cmp<dwi>_doubleword): Removed.
11077 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
11078 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
11079 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
11080 expand with cmp<dwi>_doubleword. For LTU and GEU use
11081 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
11083 * common.opt (gcolumn-info): Enable by default.
11084 * doc/invoke.texi (gcolumn-info): Document new default.
11086 2017-10-23 Richard Biener <rguenther@suse.de>
11088 PR tree-optimization/82672
11089 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
11090 Fold the stmt if we propagated into it.
11092 2017-10-23 Richard Biener <rguenther@suse.de>
11094 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
11095 (bitmap_remove_expr_from_set): ... this. All callers call this
11096 for non-constant values.
11097 (bitmap_set_subtract): Rename to...
11098 (bitmap_set_subtract_expressions): ... this. Adjust and
11100 (bitmap_set_contains_value): Remove superfluous check.
11101 (bitmap_set_replace_value): Inline into single caller ...
11102 (bitmap_value_replace_in_set): ... here and simplify.
11103 (dependent_clean): Merge into ...
11104 (clean): ... this using an overload. Adjust.
11105 (prune_clobbered_mems): Adjust.
11106 (compute_antic_aux): Likewise.
11107 (compute_partial_antic_aux): Likewise.
11109 2017-10-23 Richard Biener <rguenther@suse.de>
11111 PR tree-optimization/82129
11113 2017-08-01 Richard Biener <rguenther@suse.de>
11115 PR tree-optimization/81181
11116 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
11117 (compute_antic): ... end of iteration here.
11119 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11121 * target.def (starting_frame_offset): New hook.
11122 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
11123 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
11124 * doc/tm.texi.in: Regenerate.
11125 * hooks.h (hook_hwi_void_0): Declare.
11126 * hooks.c (hook_hwi_void_0): New function.
11127 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
11128 STARTING_FRAME_OFFSET.
11129 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
11130 * reload1.c (reload): Likewise.
11131 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
11132 instead of STARTING_FRAME_OFFSET.
11133 * function.c (try_fit_stack_local): Likewise.
11134 (assign_stack_local_1): Likewise
11135 (instantiate_virtual_regs): Likewise.
11136 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
11137 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
11138 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
11139 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
11140 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
11141 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
11142 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
11143 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
11144 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
11145 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
11146 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
11147 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
11148 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
11149 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
11150 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
11151 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
11152 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
11153 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
11154 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
11155 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
11156 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
11157 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
11158 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
11159 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
11160 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
11161 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
11162 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
11163 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
11164 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
11165 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
11166 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
11167 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
11168 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
11169 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
11170 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
11171 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
11172 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
11173 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
11174 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
11175 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
11176 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
11177 * config/avr/avr.c (avr_starting_frame_offset): Make static and
11178 return a HOST_WIDE_INT.
11179 (avr_builtin_setjmp_frame_value): Use it instead of
11180 STARTING_FRAME_OFFSET.
11181 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11182 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
11183 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
11185 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11186 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
11187 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
11188 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11189 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
11190 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
11191 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11192 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
11193 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
11194 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11195 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
11196 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
11198 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11199 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
11200 * config/mips/mips.c (mips_compute_frame_info): Refer to
11201 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11202 (mips_starting_frame_offset): New function.
11203 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11204 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
11205 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
11206 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
11207 and return a HOST_WIDE_INT.
11208 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11209 (mmix_initial_elimination_offset): Refer to
11210 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11211 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
11212 * config/pa/pa.c (pa_starting_frame_offset): New function.
11213 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
11214 (pa_expand_prologue): Likewise.
11215 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11216 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
11217 !FRAME_GROWS_DOWNWARD handling to...
11218 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11219 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
11220 !FRAME_GROWS_DOWNWARD handling to...
11221 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11222 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
11223 !FRAME_GROWS_DOWNWARD handling to...
11224 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11225 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
11227 (rs6000_starting_frame_offset): New function.
11228 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
11229 !FRAME_GROWS_DOWNWARD handling to...
11230 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11231 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
11232 !FRAME_GROWS_DOWNWARD handling to...
11233 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11234 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
11235 !FRAME_GROWS_DOWNWARD handling to...
11236 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11237 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
11238 (rs6000_starting_frame_offset): New function.
11239 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
11240 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
11241 * config/vax/vax.c (vax_starting_frame_offset): New function.
11242 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
11243 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11244 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
11245 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
11246 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11247 * system.h (STARTING_FRAME_OFFSET): Poison.
11249 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11251 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11252 SCALAR_TYPE_MODE instead of TYPE_MODE.
11254 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11255 Alan Hayward <alan.hayward@arm.com>
11256 David Sherwood <david.sherwood@arm.com>
11258 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
11260 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11261 Alan Hayward <alan.hayward@arm.com>
11262 David Sherwood <david.sherwood@arm.com>
11264 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
11266 2017-10-23 Richard Biener <rguenther@suse.de>
11268 PR tree-optimization/82129
11269 * tree-ssa-pre.c (bitmap_set_and): Remove.
11270 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
11271 canonicalizing expressions in the set to those with lowest
11272 ID rather than taking that from the first edge.
11274 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11276 * combine.c (rtx_equal_for_field_assignment_p): Use
11277 byte_lowpart_offset.
11279 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11280 Alan Hayward <alan.hayward@arm.com>
11281 David Sherwood <david.sherwood@arm.com>
11283 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
11284 to a SUBREG_PROMOTED_VAR.
11286 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11287 Alan Hayward <alan.hayward@arm.com>
11288 David Sherwood <david.sherwood@arm.com>
11290 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
11291 (expand_debug_source_expr): Likewise.
11292 * combine.c (combine_simplify_rtx): Likewise.
11293 * cse.c (fold_rtx): Likewise.
11294 * optabs.c (expand_float): Likewise.
11295 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11296 (simplify_binary_operation_1): Likewise.
11298 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11299 Alan Hayward <alan.hayward@arm.com>
11300 David Sherwood <david.sherwood@arm.com>
11302 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
11303 (record_promoted_value): Likewise.
11304 * expr.c (expand_expr_real_2): Likewise.
11305 * ree.c (update_reg_equal_equiv_notes): Likewise.
11306 (combine_set_extension): Likewise.
11307 * rtlanal.c (low_bitmask_len): Likewise.
11308 * simplify-rtx.c (neg_const_int): Likewise.
11309 (simplify_binary_operation_1): Likewise.
11311 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11312 Alan Hayward <alan.hayward@arm.com>
11313 David Sherwood <david.sherwood@arm.com>
11315 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
11316 * regcprop.c (maybe_mode_change): Likewise.
11317 * reload1.c (alter_reg): Likewise.
11319 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11321 * inchash.h (inchash::hash::add_wide_int): New function.
11322 * lto-streamer-out.c (hash_tree): Use it.
11324 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11326 * inchash.h (inchash::hash::add_wide_int): Rename to...
11327 (inchash::hash::add_hwi): ...this.
11328 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
11329 (polymorphic_call_target_hasher::hash): Likewise.
11330 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
11331 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
11332 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
11333 * lto-streamer-out.c (hash_tree): Likewise.
11334 * optc-save-gen.awk: Likewise.
11335 * tree.c (add_expr): Likewise.
11337 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11340 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
11341 for ordered inequality comparisons even with TARGET_IEEE_FP.
11343 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11346 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
11347 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11348 Expand with cmp<dwi>_doubleword.
11350 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11352 * extend.texi: Add x86 specific to 'nocf_check' attribute.
11353 List CET intrinsics.
11354 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
11355 specific to -fcf-protection option.
11357 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11359 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
11360 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
11361 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
11362 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
11363 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
11364 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
11365 (extra_objs): Add cet.o for Linux/x86 targets.
11366 (tmake_file): Add i386/t-cet for Linux/x86 targets.
11367 * config/i386/cet.c: New file.
11368 * config/i386/cetintrin.h: Likewise.
11369 * config/i386/t-cet: Likewise.
11370 * config/i386/cpuid.h (bit_SHSTK): New.
11371 (bit_IBT): Likewise.
11372 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
11373 pass IBT and SHSTK bits.
11374 * config/i386/i386-builtin-types.def
11375 (VOID_FTYPE_UNSIGNED_PVOID): New.
11376 (VOID_FTYPE_UINT64_PVOID): Likewise.
11377 * config/i386/i386-builtin.def: Add CET intrinsics.
11378 * config/i386/i386-c.c (ix86_target_macros_internal): Add
11379 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
11380 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
11381 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
11383 * config/i386/i386.c (rest_of_insert_endbranch): New.
11384 (pass_data_insert_endbranch): Likewise.
11385 (pass_insert_endbranch): Likewise.
11386 (make_pass_insert_endbranch): Likewise.
11387 (ix86_notrack_prefixed_insn_p): Likewise.
11388 (ix86_target_string): Add -mibt, -mshstk flags.
11389 (ix86_option_override_internal): Add flag_cf_protection
11391 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
11392 (ix86_print_operand): Add 'notrack' prefix output.
11393 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
11394 (ix86_expand_builtin): Expand CET intrinsics.
11395 (x86_output_mi_thunk): Add 'endbranch' instruction.
11396 * config/i386/i386.h (TARGET_IBT): New.
11397 (TARGET_IBT_P): Likewise.
11398 (TARGET_SHSTK): Likewise.
11399 (TARGET_SHSTK_P): Likewise.
11400 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
11401 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
11402 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
11403 (builtin_setjmp_setup): New pattern.
11404 (builtin_longjmp): Likewise.
11405 (rdssp<mode>): Likewise.
11406 (incssp<mode>): Likewise.
11407 (saveprevssp): Likewise.
11408 (rstorssp): Likewise.
11409 (wrss<mode>): Likewise.
11410 (wruss<mode>): Likewise.
11411 (setssbsy): Likewise.
11412 (clrssbsy): Likewise.
11413 (nop_endbr): Likewise.
11414 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
11416 * config/i386/immintrin.h: Include <cetintrin.h>.
11417 * config/i386/linux-common.h
11418 (file_end_indicate_exec_stack_and_cet): New prototype.
11419 (TARGET_ASM_FILE_END): New.
11421 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11423 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
11424 latencies instead of having separate table; make difference between
11425 integer and float costs.
11426 * i386.h (processor_costs): Remove scalar_stmt_cost,
11427 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
11428 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
11430 * x86-tune-costs.h: Remove entries which has been removed in
11431 procesor_costs from all tables; make cond_taken_branch_cost
11432 and cond_not_taken_branch_cost COST_N_INSNS based.
11434 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11436 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
11438 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11440 * config/i386/i386.md (isa): Remove fma_avx512f.
11441 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
11442 <avx512>_fmadd_<mode>_mask3<round_name>,
11443 <avx512>_fmsub_<mode>_mask<round_name>,
11444 <avx512>_fmsub_<mode>_mask3<round_name>,
11445 <avx512>_fnmadd_<mode>_mask<round_name>,
11446 <avx512>_fnmadd_<mode>_mask3<round_name>,
11447 <avx512>_fnmsub_<mode>_mask<round_name>,
11448 <avx512>_fnmsub_<mode>_mask3<round_name>,
11449 <avx512>_fmaddsub_<mode>_mask<round_name>,
11450 <avx512>_fmaddsub_<mode>_mask3<round_name>,
11451 <avx512>_fmsubadd_<mode>_mask<round_name>,
11452 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
11453 (*vec_widen_umult_even_v16si<mask_name>,
11454 *vec_widen_smult_even_v16si<mask_name>): Likewise.
11455 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
11457 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11459 * extend.texi: Add 'nocf_check' documentation.
11460 * gimple.texi: Add second parameter to
11461 gimple_build_call_from_tree.
11462 * invoke.texi: Add -fcf-protection documentation.
11463 * rtl.texi: Add REG_CALL_NOTRACK documenation.
11465 2017-10-20 Richard Biener <rguenther@suse.de>
11467 PR tree-optimization/82473
11468 * tree-vect-loop.c (vectorizable_reduction): Properly get at
11469 the largest input type.
11471 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11473 * c-attribs.c (handle_nocf_check_attribute): New function.
11474 (c_common_attribute_table): Add 'nocf_check' handling.
11475 * gimple-parser.c: Add second argument NULL to
11476 gimple_build_call_from_tree.
11477 * attrib.c (comp_type_attributes): Check nocf_check attribute.
11478 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
11480 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
11481 * common.opt: Add fcf-protection flag.
11482 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
11483 * flag-types.h: Add enum cf_protection_level.
11484 * gimple.c (gimple_build_call_from_tree): Add second parameter.
11485 Add 'nocf_check' attribute propagation to gimple call.
11486 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
11487 (gimple_build_call_from_tree): Update prototype.
11488 (gimple_call_nocf_check_p): New function.
11489 (gimple_call_set_nocf_check): Likewise.
11490 * gimplify.c: Add second argument to gimple_build_call_from_tree.
11491 * ipa-icf.c: Add nocf_check attribute in statement hash.
11492 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
11493 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
11494 * toplev.c (process_options): Add flag_cf_protection handling.
11496 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11498 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
11500 2017-10-20 Richard Biener <rguenther@suse.de>
11502 PR tree-optimization/82603
11503 * tree-if-conv.c (predicate_mem_writes): Make sure to only
11504 remove false predicated stores.
11506 2017-10-20 Richard Biener <rguenther@suse.de>
11508 * graphite-isl-ast-to-gimple.c
11509 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
11510 Remove return value and simplify, dump copied stmt after lhs
11512 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
11513 Reduce dump verbosity.
11514 (gsi_insert_earliest): Likewise.
11515 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
11516 * graphite.c (print_global_statistics): Adjust dumping.
11517 (print_graphite_scop_statistics): Likewise.
11518 (print_graphite_statistics): Do not dump loops here.
11519 (graphite_transform_loops): But here.
11521 2017-10-20 Nicolas Roche <roche@adacore.com>
11523 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
11524 * configure: Regenerate.
11526 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11529 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
11530 functions when optimizing replace GIMPLE_RETURN stmts with
11531 calls to __builtin_unreachable ().
11534 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
11535 for -fsanitize=thread link of executables.
11536 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
11537 link of executables.
11540 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
11541 New mode iterators.
11542 (<shift_insn><mode>3<mask_name>): Change the last of the 3
11543 define_insns for logical vector shifts to use VI248_AVX512BW
11544 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
11545 condition, useless isa and prefix attributes. Change the first
11546 2 of these define_insns to ...
11547 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
11548 define_insn for avx512vl.
11549 (<shift_insn><mode>3): ... and this, new define_insn without
11550 masking for non-avx512vl.
11553 * config/i386/sse.md (*andnot<mode>3,
11554 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
11555 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
11556 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
11557 not applied use empty suffix even for TARGET_AVX512VL.
11558 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
11559 is applied, supply evex,evex or evex,evex,evex instead of just
11562 2017-10-20 Julia Koval <julia.koval@intel.com>
11564 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
11565 (OPTION_MASK_ISA_GFNI_UNSET): New.
11566 (ix86_handle_option): Handle OPT_mgfni.
11567 * config/i386/cpuid.h (bit_GFNI): New.
11568 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
11569 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
11570 * config/i386/i386.c (ix86_target_string): Add -mgfni.
11571 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
11572 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
11573 * config/i386/i386.opt: Add mgfni.
11575 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
11577 * config/msp430/msp430.c (msp430_option_override): Disable
11578 -fdelete-null-pointer-checks.
11579 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
11581 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11583 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
11584 of x87 and SSE instructions.
11586 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11588 * asan.c (create_cond_insert_point): Do not update edge count.
11589 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
11590 (afdo_propagate_circuit): Likewise.
11591 (afdo_calculate_branch_prob): Likewise.
11592 (afdo_annotate_cfg): Likewise.
11593 * basic-block.h (struct edge_def): Remove count.
11594 (edge_def::count): New accessor.
11595 * bb-reorder.c (rotate_loop): Update.
11596 (find_traces_1_round): Update.
11597 (connect_traces): Update.
11598 (sanitize_hot_paths): Update.
11599 * cfg.c (unchecked_make_edge): Update.
11600 (make_single_succ_edge): Update.
11601 (check_bb_profile): Update.
11602 (dump_edge_info): Update.
11603 (update_bb_profile_for_threading): Update.
11604 (scale_bbs_frequencies_int): Update.
11605 (scale_bbs_frequencies_gcov_type): Update.
11606 (scale_bbs_frequencies_profile_count): Update.
11607 (scale_bbs_frequencies): Update.
11608 * cfganal.c (connect_infinite_loops_to_exit): Update.
11609 * cfgbuild.c (compute_outgoing_frequencies): Update.
11610 (find_many_sub_basic_blocks): Update.
11611 * cfgcleanup.c (try_forward_edges): Update.
11612 (try_crossjump_to_edge): Update
11613 * cfgexpand.c (expand_gimple_cond): Update
11614 (expand_gimple_tailcall): Update
11615 (construct_exit_block): Update
11616 * cfghooks.c (verify_flow_info): Update
11617 (redirect_edge_succ_nodup): Update
11618 (split_edge): Update
11619 (make_forwarder_block): Update
11620 (duplicate_block): Update
11621 (account_profile_record): Update
11622 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11623 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11624 * cfgloopmanip.c (scale_loop_profile): Update.
11626 (lv_adjust_loop_entry_edge): Update.
11627 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11628 (force_nonfallthru_and_redirect): Update.
11629 (purge_dead_edges): Update.
11630 (rtl_flow_call_edges_add): Update.
11631 * cgraphunit.c (init_lowered_empty_function): Update.
11632 (cgraph_node::expand_thunk): Update.
11633 * gimple-pretty-print.c (dump_probability): Update.
11634 (dump_edge_probability): Update.
11635 * gimple-ssa-isolate-paths.c (isolate_path): Update.
11636 * haifa-sched.c (sched_create_recovery_edges): Update.
11637 * hsa-gen.c (convert_switch_statements): Update.
11638 * ifcvt.c (dead_or_predicable): Update.
11639 * ipa-inline-transform.c (inline_transform): Update.
11640 * ipa-split.c (split_function): Update.
11641 * ipa-utils.c (ipa_merge_profiles): Update.
11642 * loop-doloop.c (add_test): Update.
11643 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
11644 * lto-streamer-in.c (input_cfg): Update.
11645 (input_function): Update.
11646 * lto-streamer-out.c (output_cfg): Update.
11647 * modulo-sched.c (sms_schedule): Update.
11648 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11649 * predict.c (maybe_hot_edge_p): Update.
11650 (unlikely_executed_edge_p): Update.
11651 (probably_never_executed_edge_p): Update.
11652 (dump_prediction): Update.
11653 (drop_profile): Update.
11654 (propagate_unlikely_bbs_forward): Update.
11655 (determine_unlikely_bbs): Update.
11656 (force_edge_cold): Update.
11657 * profile.c (compute_branch_probabilities): Update.
11658 * reg-stack.c (better_edge): Update.
11659 * shrink-wrap.c (handle_simple_exit): Update.
11660 * tracer.c (better_p): Update.
11661 * trans-mem.c (expand_transaction): Update.
11662 (split_bb_make_tm_edge): Update.
11663 * tree-call-cdce.c: Update.
11664 * tree-cfg.c (gimple_find_sub_bbs): Update.
11665 (gimple_split_edge): Update.
11666 (gimple_duplicate_sese_region): Update.
11667 (gimple_duplicate_sese_tail): Update.
11668 (gimple_flow_call_edges_add): Update.
11669 (insert_cond_bb): Update.
11670 (execute_fixup_cfg): Update.
11671 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
11672 * tree-complex.c (expand_complex_div_wide): Update.
11673 * tree-eh.c (lower_resx): Update.
11674 (unsplit_eh): Update.
11675 (cleanup_empty_eh_move_lp): Update.
11676 * tree-inline.c (copy_edges_for_bb): Update.
11677 (freqs_to_counts): Update.
11678 (copy_cfg_body): Update.
11679 * tree-ssa-dce.c (remove_dead_stmt): Update.
11680 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11681 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11682 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11683 (unloop_loops): Update.
11684 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11685 * tree-ssa-loop-split.c (connect_loops): Update.
11686 (split_loop): Update.
11687 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11688 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11689 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11690 * tree-ssa-reassoc.c (branch_fixup): Update.
11691 * tree-ssa-tail-merge.c (replace_block_by): Update.
11692 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11693 (compute_path_counts): Update.
11694 (update_profile): Update.
11695 (recompute_probabilities): Update.
11696 (update_joiner_offpath_counts): Update.
11697 (estimated_freqs_path): Update.
11698 (freqs_to_counts_path): Update.
11699 (clear_counts_path): Update.
11700 (ssa_fix_duplicate_block_edges): Update.
11701 (duplicate_thread_path): Update.
11702 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11703 (case_bit_test_cmp): Update.
11704 (collect_switch_conv_info): Update.
11705 (gen_inbound_check): Update.
11706 (do_jump_if_equal): Update.
11707 (emit_cmp_and_jump_insns): Update.
11708 * tree-tailcall.c (decrease_profile): Update.
11709 (eliminate_tail_call): Update.
11710 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11711 (vect_do_peeling): Update.
11712 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11713 * ubsan.c (ubsan_expand_null_ifn): Update.
11714 (ubsan_expand_ptr_ifn): Update.
11715 * value-prof.c (gimple_divmod_fixed_value): Update.
11716 (gimple_mod_pow2): Update.
11717 (gimple_mod_subtract): Update.
11718 (gimple_ic): Update.
11719 (gimple_stringop_fixed_value): Update.
11721 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11724 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
11726 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
11728 PR rtl-optimization/82395
11729 * ira-color.c (allocno_priority_compare_func): Fix comparison step
11730 based on non_spilled_static_chain_regno_p.
11732 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11734 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
11735 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
11736 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
11738 2017-10-19 Martin Sebor <msebor@redhat.com>
11740 PR tree-optimization/82596
11741 * tree.c (array_at_struct_end_p): Handle STRING_CST.
11743 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11745 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
11746 (get_mem_refs_of_builtin_call): Likewise.
11747 * builtins.c (expand_builtin_apply): Adjust call to
11748 allocate_dynamic_stack_space.
11749 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
11750 the third argument to allocate_dynamic_stack_space, otherwise -1.
11751 (expand_builtin): Deal with all alloca variants.
11752 (is_inexpensive_builtin): Likewise.
11753 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
11754 * calls.c (special_function_p): Deal with all alloca variants.
11755 (initialize_argument_information): Adjust call to
11756 allocate_dynamic_stack_space.
11757 (expand_call): Likewise.
11758 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
11759 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
11760 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
11761 use it for the stack usage computation.
11762 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
11763 * function.c (gimplify_parameters): Call build_alloca_call_expr.
11764 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
11765 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
11766 (in_loop_p): Remove first argument and useless check.
11767 (pass_walloca::execute): Remove useless test and adjust call to above.
11768 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
11769 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
11770 (gimplify_call_expr): Deal with all alloca variants.
11771 * hsa-gen.c (gen_hsa_alloca): Likewise.
11772 (gen_hsa_insns_for_call): Likewise.
11773 * ipa-pure-const.c (special_builtin_state): Likewise.
11774 * tree-chkp.c (chkp_build_returned_bound): Likewise.
11775 * tree-object-size.c (alloc_object_size): Likewise.
11776 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
11777 (call_may_clobber_ref_p_1): Likewise.
11778 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
11779 (ccp_fold_stmt): Likewise.
11780 (optimize_stack_restore): Likewise.
11781 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
11782 (mark_all_reaching_defs_necessary_1): Likewise.
11783 (propagate_necessity): Likewise.
11784 (eliminate_unnecessary_stmts): Likewise.
11785 * tree.c (build_common_builtin_nodes): Build
11786 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
11787 (build_alloca_call_expr): New function.
11788 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
11789 (CASE_BUILT_IN_ALLOCA): Likewise.
11790 (build_alloca_call_expr): Declare.
11791 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
11793 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11796 * dwarf2out.c (new_die_raw): New static inline function.
11797 (new_die): Use it to create the DIE.
11798 (add_AT_external_die_ref): Likewise.
11799 (clone_die): Likewise.
11800 (clone_as_declaration): Likewise.
11801 (dwarf2out_vms_debug_main_pointer): Likewise.
11802 (base_type_die): Likewise. Remove early return for corner cases.
11803 Do not call add_pubtype on the DIE here.
11804 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
11805 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11806 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11807 native order exists for base types, attach the DIE manually and call
11808 add_pubtype on it. Do not equate a reverse order DIE to the type.
11810 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11812 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11813 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11815 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11817 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11819 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11820 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11821 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11822 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11823 * opts.c (sanitizer_opts): Add builtin.
11824 * ubsan.c (instrument_builtin): New function.
11825 (pass_ubsan::execute): Call it.
11826 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11827 * doc/invoke.texi: Document -fsanitize=builtin.
11829 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11830 builtins, store max (log2 (align), 0) into uchar field instead of
11831 align into uptr field.
11832 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11833 store uchar 0 field instead of uptr 0 field.
11834 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11835 instead of passing one address of struct with 2 locations pass
11836 two addresses of structs with 1 location each.
11837 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11838 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11839 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11840 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11841 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11842 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11843 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
11844 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
11846 2017-10-19 Martin Liska <mliska@suse.cz>
11849 * file-find.c (remove_prefix): Remove.
11850 * file-find.h (remove_prefix): Likewise.
11851 * gcc-ar.c: Remove smartness of lookup.
11853 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11855 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11856 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11857 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11859 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11862 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11863 (setcc + and to xor + setcc): New peephole2.
11865 2017-10-19 Tom de Vries <tom@codesourcery.com>
11867 * doc/sourcebuild.texi (Test Directives, Variants of
11868 dg-require-support): Add dg-require-stack-size.
11870 2017-10-19 Martin Liska <mliska@suse.cz>
11873 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11874 that have a large alignment.
11875 (gimplify_target_expr): Likewise.
11877 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11879 PR rtl-optimization/82602
11880 * ira.c (rtx_moveable_p): Return false for volatile asm.
11882 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11885 * config/i386/i386-modes.def (CCGZ): New CC mode.
11886 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11887 * config/i386/predicates.md (ix86_comparison_operator):
11889 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11890 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11891 with double-word subtraction.
11892 (put_condition_code): Handle CCGZmode.
11894 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11896 * wide-int.cc (debug (const wide_int &)): New.
11897 (debug (const wide_int *)): New.
11898 (debug (const widest_int &)): New.
11899 (debug (const widest_int *)): New.
11901 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11903 PR middle-end/82556
11904 * lra-constraints.c (curr_insn_transform): Use non-input operand
11905 instead of output one for matched reload.
11907 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11909 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11910 (tree-ssa-loop-ivopts.h): New header file.
11911 (struct builtin_info): New fields.
11912 (classify_builtin_1): Compute and record base and offset parts for
11913 memset builtin partition by calling strip_offset.
11914 (offset_cmp, fuse_memset_builtins): New functions.
11915 (finalize_partitions): Fuse adjacent memset partitions by calling
11917 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
11918 Expose the interface.
11919 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
11921 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11923 PR tree-optimization/82574
11924 * tree-loop-distribution.c (find_single_drs): New parameter. Check
11925 that data reference must be executed exactly once per iteration
11926 against the outermost loop in nest.
11927 (classify_partition): Update call to above function.
11929 2017-10-18 Richard Biener <rguenther@suse.de>
11931 PR tree-optimization/82591
11932 * graphite.c (graphite_transform_loops): Move code gen message
11934 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11935 Here. Handle scop_to_isl_ast failing.
11936 (scop_to_isl_ast): Limit the number of ISL operations.
11938 2017-10-18 Richard Biener <rguenther@suse.de>
11940 * graphite-isl-ast-to-gimple.c
11941 (translate_isl_ast_to_gimple::set_rename): Simplify.
11942 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
11943 (graphite_copy_stmts_from_block): ... here.
11944 (copy_bb_and_scalar_dependences): Simplify.
11945 (add_parameters_to_ivs_params): Canonicalize.
11946 (generate_entry_out_of_ssa_copies): Simplify.
11947 * graphite-sese-to-poly.c (extract_affine_name): Simplify
11948 by passing in ISL dimension.
11949 (parameter_index_in_region_1): Rename to ...
11950 (parameter_index_in_region): ... this.
11951 (extract_affine): Adjust assert, pass down parameter index.
11952 (add_param_constraints): Use range-info when available.
11953 (build_scop_context): Adjust.
11954 * sese.c (new_sese_info): Adjust.
11955 (free_sese_info): Likewise.
11956 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
11957 Remove unused typedefs.
11958 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
11960 2017-10-18 Martin Liska <mliska@suse.cz>
11962 * combine.c (simplify_compare_const): Add gcc_fallthrough.
11964 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11966 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
11967 (s390_sched_init): Do not reset s390_sched_state if we entered the
11968 current basic block via a fallthru edge and all others are unlikely.
11970 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11972 * config/s390/s390.c (NUM_SIDES): New variable.
11973 (LONGRUNNING_THRESHOLD): New variable.
11974 (LATENCY_FACTOR): New variable.
11975 (s390_sched_score): Decrease score for long-running instructions on
11977 (s390_sched_variable_issue): Perform bookkeeping for long-running
11980 2017-10-18 Richard Biener <rguenther@suse.de>
11982 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
11983 Simplify with removal of the parameter rename map.
11984 (set_rename): Likewise.
11985 (should_copy_to_new_region): Likewise.
11986 (graphite_copy_stmts_from_block): Likewise.
11987 (copy_bb_and_scalar_dependences): Remove initialization of
11988 unused copied_bb_map.
11989 (copy_def): Remove.
11990 (copy_internal_parameters): Likewise.
11991 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
11992 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
11993 Use INTEGRAL_TYPE_P.
11994 (parameter_index_in_region_1): Rename to ...
11995 (assign_parameter_index_in_region): ... this. Assert we have
11996 a parameter we handle.
11997 (scan_tree_for_params): Adjust.
11998 * sese.h (parameter_rename_map_t): Remove.
11999 (struct sese_info_t): Remove unused parameter_rename_map and
12000 copied_bb_map members.
12001 * sese.c (new_sese_info): Adjust.
12002 (free_sese_info): Likewise.
12004 2017-10-18 Martin Liska <mliska@suse.cz>
12007 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
12008 on an abnormal edge.
12010 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
12012 * doc/invoke.texi (ffunction-sections and fdata-sections):
12015 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
12017 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
12018 the use statement can throw internally.
12020 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
12022 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
12023 any RTX present on the RHS of a SET.
12024 * compare-elim.c (try_eliminate_compare): Restore comment.
12026 2017-10-17 Jakub Jelinek <jakub@redhat.com>
12028 * langhooks.h (struct lang_hooks): Document that tree_size langhook
12029 may be also called on tcc_type nodes.
12030 * langhooks.c (lhd_tree_size): Likewise.
12032 2017-10-17 David Malcolm <dmalcolm@redhat.com>
12034 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
12035 format_warning_at_substring.
12036 (maybe_warn): Convert source_range * param to a location_t. Pass
12037 UNKNOWN_LOCATION rather than NULL to fmtwarn.
12038 (format_directive): Remove code to extract source_ranges and
12039 source_range * in favor of just a location_t.
12040 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
12042 * substring-locations.c (format_warning_va): Convert
12043 source_range * param to a location_t.
12044 (format_warning_at_substring): Likewise.
12045 * substring-locations.h (format_warning_va): Likewise.
12046 (format_warning_at_substring): Likewise.
12048 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
12050 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
12052 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
12053 and scatter/gather ops.
12055 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
12056 vec_gather_load and vec_scatter_store.
12057 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
12058 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
12059 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
12060 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
12061 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
12062 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
12064 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
12066 * reg-stack.c (compare_for_stack_reg): Add bool argument.
12067 Detect FTST instruction and handle its register pops. Only pop
12068 second operand if can_pop_second_op is true.
12069 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
12070 set can_pop_second_op to false in the compare_for_stack_reg call.
12072 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
12073 output_fp_compare for stack register operands.
12074 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
12075 instructions here. Do not emit stack register pops here. Assert
12076 that FCOMPP pops next to top stack register. Rewrite function.
12078 2017-10-17 Nathan Sidwell <nathan@acm.org>
12080 PR middle-end/82577
12081 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
12082 use DECL_ASSEMBLER_NAME_RAW.
12084 PR middle-end/82546
12085 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
12088 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
12089 Wilco Dijkstra <wilco.dijkstra@arm.com>
12091 * builtins.c (expand_builtin_update_setjmp_buf): Add a
12092 converstion to Pmode from the buf_addr.
12094 2017-10-17 Richard Biener <rguenther@suse.de>
12096 * graphite-dependences.c (scop_get_reads_and_writes): Change
12097 output parameters to references.
12099 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
12101 PR 71026/tree-optimization
12102 * fold-const.c (distribute_real_division): Removed.
12103 (fold_binary_loc): Remove calls to distribute_real_divison.
12105 2017-10-17 Richard Biener <rguenther@suse.de>
12107 * graphite-scop-detection.c
12108 (scop_detection::stmt_has_simple_data_refs_p): Always use
12109 the full nest as region.
12110 (try_generate_gimple_bb): Likewise.
12111 * sese.c (scalar_evolution_in_region): Simplify now that
12112 SCEV can handle instantiation in regions.
12113 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
12114 in the non-loop part of a function if requested.
12116 2017-10-17 Richard Biener <rguenther@suse.de>
12118 PR tree-optimization/82563
12119 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
12121 (graphite_regenerate_ast_isl): Call it.
12122 * graphite-scop-detection.c (build_scops): Remove entry edge split.
12124 2017-10-17 Jakub Jelinek <jakub@redhat.com>
12126 PR tree-optimization/82549
12127 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
12128 Formatting fixes. Instead of calling make_bit_field_ref with negative
12131 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
12133 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
12134 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
12135 _mm_maskz_reduce_ss): New.
12136 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
12137 __builtin_ia32_reducess_mask): Ditto..
12138 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
12139 * config/i386/sse.md (reduces<mode>): Renamed to ...
12140 (reduces<mode><mask_scalar_name>): ... this.
12141 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
12143 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
12144 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
12146 2017-10-16 David Malcolm <dmalcolm@redhat.com>
12148 * Makefile.in (OBJS): Add unique-ptr-tests.o.
12149 * selftest-run-tests.c (selftest::run_tests): Call
12150 selftest::unique_ptr_tests_cc_tests.
12151 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
12152 * unique-ptr-tests.cc: New file.
12154 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
12157 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12159 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12160 (make_hard_regno_born, make_hard_regno_dead): Update
12161 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
12163 2017-10-16 Jeff Law <law@redhat.com>
12165 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12167 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
12169 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
12171 2017-10-16 Olivier Hainque <hainque@adacore.com>
12173 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
12174 with_cpu if we were configured for an e500v2 target cpu name.
12176 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
12178 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
12179 * doc/invoke.texi: Document +nodsp as a valid extension for
12182 2017-10-16 Martin Liska <mliska@suse.cz>
12184 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
12185 (test_set_range): Likewise.
12186 (test_range_functions): Rename to ...
12187 (test_bit_in_range): ... this.
12188 (sbitmap_c_tests): Add new test.
12190 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12192 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
12194 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
12195 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
12197 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12199 * config/aarch64/aarch64-builtins.c
12200 (aarch64_types_quadopu_lane_qualifiers): New.
12201 (TYPES_QUADOPU_LANE): New.
12202 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
12203 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
12204 (aarch64_<sur>dot_laneq<vsi2qi>): New.
12205 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
12206 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
12207 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
12208 (Vdottype, DOTPROD): New.
12209 (sur): Add SDOT and UDOT.
12211 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12213 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
12214 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
12215 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
12216 Add TARGET_DOTPROD.
12217 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
12218 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
12219 Enable TARGET_DOTPROD.
12220 (cortex-a75.cortex-a55): Likewise.
12221 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
12223 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12225 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
12226 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
12228 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
12230 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
12231 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
12232 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
12233 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
12234 * config/arm/types.md (neon_dot, neon_dot_q): New.
12235 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
12237 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12239 * config/arm/arm.h (TARGET_DOTPROD): New.
12240 * config/arm/arm.c (arm_arch_dotprod): New.
12241 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
12242 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
12243 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
12244 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
12245 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
12246 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
12247 * doc/invoke.texi (armv8.2-a): Document dotprod
12249 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12251 * i386.c (ix86_vec_cost): New function.
12252 (ix86_rtx_costs): Handle vector operations better.
12253 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
12254 * x86-tune-costs.h: Add new costs to all tables.
12256 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12258 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
12260 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
12261 divsd, sqrtss and sqrtsd
12262 * x86-tune-costs.h: Add new entries to all costs.
12263 (znver1_cost): Fix to match real instruction latencies.
12265 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12266 Michael Collison <michael.collison@arm.com>
12268 * compare-elim.c: Include emit-rtl.h.
12269 (can_merge_compare_into_arith): New function.
12270 (try_validate_parallel): Likewise.
12271 (try_merge_compare): Likewise.
12272 (try_eliminate_compare): Call the above when no previous clobber
12274 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
12277 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12279 PR middle-end/62263
12280 PR middle-end/82498
12281 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
12282 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
12284 PR middle-end/62263
12285 PR middle-end/82498
12286 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
12287 to be any operand_equal_p operands. For & (B - 1) require
12288 B to be power of 2. Recognize
12289 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
12291 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12294 * optabs.c (expand_memory_blockage): Fix call of
12295 targetm.have_memory_blockage.
12297 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12300 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
12301 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
12302 objects to extra_objs instead of overwriting it.
12304 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12306 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
12307 Use any_fp_register_operand as operand[3] predicate. Simplify
12308 equality test for operands[2] and operands[4] memory location.
12309 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
12310 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
12311 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
12312 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
12313 any_fp_register_operand as operand[1] predicate. Simplify
12314 equality test for operands[0] and operands[3] memory location.
12315 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
12316 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
12317 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
12319 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12321 * target-insns.def: Add memory_blockage.
12322 * optabs.c (expand_memory_blockage): New function.
12323 (expand_asm_memory_barrier): Rename ...
12324 (expand_asm_memory_blockage): ... to this.
12325 (expand_mem_thread_fence): Call expand_memory_blockage
12326 instead of expand_asm_memory_barrier.
12327 (expand_mem_singnal_fence): Ditto.
12328 (expand_atomic_load): Ditto.
12329 (expand_atomic_store): Ditto.
12330 * doc/md.texi (Standard Pattern Names For Generation):
12331 Document memory_blockage instruction pattern.
12333 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
12335 * config/rl78/rl78.c (rl78_emit_libcall): New function.
12336 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
12337 * config/rl78/rl78.md: New define_expand "adddi3".
12339 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12341 * cfghooks.c (verify_flow_info): Disable check that all probabilities
12344 2017-10-13 Jeff Law <law@redhat.com>
12346 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
12348 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12351 * internal-fn.c (expand_mul_overflow): If both operands have
12352 the same highpart of -1 or 0 and the topmost bit of lowpart
12353 is different, overflow is if res <= 0 rather than res < 0.
12355 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
12357 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
12358 TARGET_P9_VECTOR code for unaligned_load case.
12360 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12362 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
12364 2017-10-13 Nathan Sidwell <nathan@acm.org>
12366 * tree-core.h (tree_contains_struct): Make bool.
12367 * tree.c (tree_contains_struct): Likewise.
12368 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
12369 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
12370 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
12371 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
12372 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
12374 2017-10-13 Richard Biener <rguenther@suse.de>
12376 * graphite-isl-ast-to-gimple.c
12377 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
12378 parameters and dominance check.
12379 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
12380 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
12381 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
12382 Do not update SSA form here or do intermediate IL verification.
12383 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
12384 (graphite_initialize): Remove check on the number of loops in
12385 the function and inline into graphite_transform_loops.
12386 (graphite_finalize): Inline into graphite_transform_loops.
12387 (graphite_transform_loops): Perform SSA update and IL verification
12389 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
12391 2017-10-13 Richard Biener <rguenther@suse.de>
12393 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
12394 graphite_expression_type_precision): Avoid global constructor
12396 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
12397 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
12398 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
12399 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
12400 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
12401 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
12403 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
12406 * config/i386/i386.h (ix86_red_zone_size): New.
12407 * config/i386/i386.md (push peephole2s): Replace
12408 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
12410 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12411 Alan Hayward <alan.hayward@arm.com>
12412 David Sherwood <david.sherwood@arm.com>
12414 * combine.c (can_change_dest_mode): Reject changes in
12415 REGMODE_NATURAL_SIZE.
12417 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12418 Alan Hayward <alan.hayward@arm.com>
12419 David Sherwood <david.sherwood@arm.com>
12421 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
12422 (expand_debug_source_expr): Likewise.
12423 * combine.c (combine_simplify_rtx): Likewise.
12424 * cse.c (fold_rtx): Likewise.
12425 * fwprop.c (canonicalize_address): Likewise.
12426 * targhooks.c (default_shift_truncation_mask): Likewise.
12428 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12429 Alan Hayward <alan.hayward@arm.com>
12430 David Sherwood <david.sherwood@arm.com>
12432 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
12433 (widened_mode): Likewise.
12434 (expand_unop): Likewise.
12435 * ree.c (transform_ifelse): Likewise.
12436 (merge_def_and_ext): Likewise.
12437 (combine_reaching_defs): Likewise.
12438 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12440 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12441 Alan Hayward <alan.hayward@arm.com>
12442 David Sherwood <david.sherwood@arm.com>
12444 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
12445 * combine.c (gen_lowpart_for_combine): Likewise.
12446 * dwarf2out.c (rtl_for_decl_location): Likewise.
12447 * final.c (alter_subreg): Likewise.
12448 * rtlhooks.c (gen_lowpart_general): Likewise.
12449 (gen_lowpart_if_possible): Likewise.
12451 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12452 Alan Hayward <alan.hayward@arm.com>
12453 David Sherwood <david.sherwood@arm.com>
12455 * calls.c (expand_call): Use subreg_lowpart_offset.
12456 * cse.c (cse_insn): Likewise.
12457 * regcprop.c (copy_value): Likewise.
12458 (copyprop_hardreg_forward_1): Likewise.
12460 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12463 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
12464 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
12465 =Q constraints to +Q and into insn condition add check
12466 that operands[0] and operands[1] are equal.
12467 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
12468 =Q constraints to +Q and into insn condition add check
12469 that operands[0] is equal to either operands[1] or operands[2].
12472 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
12473 instead of handling MINUS_EXPR twice (once for each argument),
12474 canonicalize operand order and handle just once, use rtype where
12475 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
12478 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
12479 any values of __C while still being pattern recognizable as a simple
12480 rotate instruction.
12482 2017-10-13 Richard Biener <rguenther@suse.de>
12484 PR tree-optimization/82451
12486 2017-10-02 Richard Biener <rguenther@suse.de>
12488 PR tree-optimization/82355
12489 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12490 a mapping for the enclosing loop but avoid generating one for
12491 the loop tree root.
12492 (copy_bb_and_scalar_dependences): Remove premature codegen
12493 error on PHIs in blocks duplicated into multiple places.
12494 * graphite-scop-detection.c
12495 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12496 in the region use it as loop and nest to analyze the DR in.
12497 (try_generate_gimple_bb): Likewise.
12498 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12499 (add_loop_constraints): For blocks in a loop not in the region
12500 create a dimension with a single iteration.
12501 * sese.h (gbb_loop_at_index): Remove assert.
12503 * cfgloop.c (loop_preheader_edge): For the loop tree root
12504 return the single successor of the entry block.
12505 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12506 Reset the SCEV hashtable and niters.
12507 * graphite-scop-detection.c
12508 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
12509 assert that we only have POLYNOMIAL_CHREC that vary in loops
12510 contained in the region.
12511 (scop_detection::graphite_can_represent_expr): Adjust.
12512 (scop_detection::stmt_has_simple_data_refs_p): For loops
12513 not in the region set loop to NULL. The nest is now the
12514 entry edge to the region.
12515 (try_generate_gimple_bb): Likewise.
12516 * sese.c (scalar_evolution_in_region): Adjust for
12517 instantiate_scev change.
12518 * tree-data-ref.h (graphite_find_data_references_in_stmt):
12519 Make nest parameter the edge into the region.
12520 (create_data_ref): Likewise.
12521 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
12522 entry edge into a region and adjust instantiate_scev calls.
12523 (create_data_ref): Likewise.
12524 (graphite_find_data_references_in_stmt): Likewise.
12525 (find_data_references_in_stmt): Pass the loop preheader edge
12526 from the nest argument.
12527 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
12528 parameter the edge into the region.
12529 (instantiate_parameters): Use the loop preheader edge as entry.
12530 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
12532 (get_instantiated_value_entry): Make instantiate_below parameter
12533 the edge into the region.
12534 (instantiate_scev_name): Likewise. Adjust dominance checks,
12535 when we cannot use loop-based instantiation instantiate by
12536 walking use-def chains.
12537 (instantiate_scev_poly): Adjust.
12538 (instantiate_scev_binary): Likewise.
12539 (instantiate_scev_convert): Likewise.
12540 (instantiate_scev_not): Likewise.
12541 (instantiate_array_ref): Remove.
12542 (instantiate_scev_3): Likewise.
12543 (instantiate_scev_2): Likewise.
12544 (instantiate_scev_1): Likewise.
12545 (instantiate_scev_r): Do not blindly handle N-operand trees.
12546 Do not instantiate array-refs. Handle all constants and invariants.
12547 (instantiate_scev): Make instantiate_below parameter
12548 the edge into the region.
12549 (resolve_mixers): Use the loop preheader edge for the region
12550 parameter to instantiate_scev_r.
12551 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
12553 2017-10-13 Richard Biener <rguenther@suse.de>
12555 PR tree-optimization/82525
12556 * graphite-isl-ast-to-gimple.c
12557 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
12559 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
12560 Fail code generation when we cannot represent the isl integer.
12561 (binary_op_to_tree): Elide modulo operations that are no-ops
12562 in the type we code generate. Remove now superfluous code
12565 2017-10-13 Richard Biener <rguenther@suse.de>
12567 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
12568 (scop_detection::harmful_loop_in_region): Remove premature
12569 IV type restriction.
12570 (scop_detection::graphite_can_represent_scev): We can handle
12571 pointer IVs just fine.
12573 2017-10-13 Alan Modra <amodra@gmail.com>
12575 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
12576 "Clobbers and Scratch Registers". Add paragraph on
12577 alternative to clobbers for scratch registers and OpenBLAS
12580 2017-10-13 Alan Modra <amodra@gmail.com>
12582 * doc/extend.texi (Clobbers): Correct vax example. Delete old
12583 example of a memory input for a string of known length. Move
12584 commentary out of table. Add a number of new examples
12585 covering array memory inputs.
12587 2017-10-12 Martin Liska <mliska@suse.cz>
12589 PR tree-optimization/82493
12590 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
12591 (test_range_functions): New function.
12592 (sbitmap_c_tests): Likewise.
12593 * selftest-run-tests.c (selftest::run_tests): Run new tests.
12594 * selftest.h (sbitmap_c_tests): New function.
12596 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12598 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12600 * config/rs6000/amo.h: Fix spacing issue.
12602 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12605 * config/i386/i386.md (*ashl<mode>3_mask_1,
12606 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
12607 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
12610 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12612 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
12613 (profile_probability): Set max_probability
12614 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
12616 * profile-count.c (profile_probability::differs_from_p): Do not
12617 rely on max_probaiblity == 10000
12619 2017-10-12 Jeff Law <law@redhat.com>
12621 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
12624 2017-10-12 Martin Sebor <msebor@redhat.com>
12628 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
12629 (handle_alias_pairs): Call it.
12630 * common.opt (-Wattribute-alias): New option.
12631 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
12632 * doc/invoke.texi (-Wattribute-alias): Document.
12634 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
12637 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12639 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12641 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12642 (make_hard_regno_born, make_hard_regno_dead): Update
12643 bb_killed_pseudos and bb_gen_pseudos.
12645 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12647 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
12649 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
12651 * config/alpha/alpha.c (alpha_split_conditional_move):
12652 Use std::swap instead of manually swapping.
12653 (alpha_stdarg_optimize_hook): Ditto.
12654 (alpha_canonicalize_comparison): Ditto.
12656 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12658 * tree-loop-distribution.c (struct builtin_info): New struct.
12659 (struct partition): Refactor fields into struct builtin_info.
12660 (partition_free): Free struct builtin_info.
12661 (build_size_arg_loc, build_addr_arg_loc): Delete.
12662 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
12663 information from struct builtin_info.
12664 (find_single_drs): New function refactored from classify_partition.
12665 Also moved builtin validity checks to this function.
12666 (compute_access_range, alloc_builtin): New functions.
12667 (classify_builtin_st, classify_builtin_ldst): New functions.
12668 (classify_partition): Refactor code into functions find_single_drs,
12669 classify_builtin_st and classify_builtin_ldst.
12670 (distribute_loop): Don't do runtime alias check when distributing
12672 (find_seed_stmts_for_distribution): New function.
12673 (pass_loop_distribution::execute): Refactor code finding seed
12674 stmts into above function. Support distribution for the innermost
12675 two-level loop nest. Adjust dump information.
12677 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12679 * tree-loop-distribution.c: Adjust the general comment.
12680 (NUM_PARTITION_THRESHOLD): New macro.
12681 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
12682 (classify_partition): Skip builtin pattern of loop nest's inner loop.
12683 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
12684 in call to build_partition_graph.
12685 (finalize_partitions): New parameter. Make loop distribution more
12686 conservative by fusing more partitions.
12687 (distribute_loop): Don't do runtime alias check in case of loop nest
12689 (find_seed_stmts_for_distribution): New function.
12690 (prepare_perfect_loop_nest): New function.
12691 (pass_loop_distribution::execute): Refactor code finding seed stmts
12692 and loop nest into above functions. Support loop nest distribution.
12693 Adjust dump information accordingly.
12695 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12697 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
12698 and set PTYPE_SEQUENTIAL for merged partition.
12700 2017-10-12 Richard Biener <rguenther@suse.de>
12702 PR tree-optimization/69728
12704 2017-09-19 Richard Biener <rguenther@suse.de>
12706 PR tree-optimization/69728
12707 * graphite-sese-to-poly.c (schedule_error): New global.
12708 (add_loop_schedule): Handle empty domain by failing the
12710 (build_original_schedule): Handle schedule_error.
12712 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
12713 domain by returning an unchanged schedule.
12715 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12717 * genrecog.c (validate_pattern): For VEC_SELECT verify that
12718 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
12720 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
12722 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
12725 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12728 * expr.c (store_field): Don't optimize away bitsize == 0 store
12729 from CALL_EXPR with addressable return type.
12731 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12733 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
12734 * config/rs6000/rs6000.md (sel): Delete mode attribute.
12735 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
12736 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
12737 TARGET_ISEL instead of TARGET_ISEL<sel>.
12739 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
12741 * config/rs6000/rs6000.c
12742 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
12744 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12746 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
12748 (reg_or_cint_operand): Fix comment.
12749 (reg_or_zero_operand): New predicate.
12750 * config/rs6000/rs6000-protos.h (output_isel): Delete.
12751 * config/rs6000/rs6000.c (output_isel): Delete.
12752 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
12753 instead of reg_or_cint_operand. Output instruction directly (not via
12755 (isel_unsigned_<mode>): Ditto.
12756 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
12757 gpc_reg_operand. Add an instruction alternative for this. Output
12758 instruction directly.
12759 (*isel_reversed_unsigned_<mode>): Ditto.
12761 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12763 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
12764 (TARGET_CANONICALIZE_COMPARISON): Define.
12766 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
12769 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
12770 Check whether the dest is REG before adding REG_EQUIV note.
12772 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12775 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12777 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12778 (make_hard_regno_born, make_hard_regno_dead): Update
12779 bb_killed_pseudos and bb_gen_pseudos.
12781 2017-10-11 Nathan Sidwell <nathan@acm.org>
12783 * incpath.h (enum incpath_kind): Name enum, prefix values.
12784 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
12785 * incpath.c (heads, tails): Use INC_MAX.
12786 (add_env_var_paths, add_standard_paths): Use incpath_kind.
12787 (merge_include_chains, split_quote_chain,
12788 register_include_chains): Update incpath_kind names.
12789 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
12790 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
12792 (add_framework_path, darwin_register_objc_includes): Likewise.
12793 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
12795 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12797 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
12798 Do not use float_operator operator predicate.
12799 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
12800 * config/i386/predicates.md (float_operator): Remove predicate.
12802 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12804 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
12805 (*jccxf_i387): Ditto.
12806 (*jcc<mode>_i387): Ditto.
12807 (*jccu<mode>_i387): Ditto.
12808 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12809 (*jcc_*_i387 splitters): Remove.
12810 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12811 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12812 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12815 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12817 * profile-count.h (slow_safe_scale_64bit): New function.
12818 (safe_scale_64bit): New inline.
12819 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12820 * profile-count.c: Include wide-int.h
12821 (slow_safe_scale_64bit): New.
12823 2017-10-11 Nathan Sidwell <nathan@acm.org>
12825 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12826 HAS_DECL_ASSEMBLER_NAME_P.
12827 * gimple-expr.c (gimple_decl_printable_name: Check
12828 HAS_DECL_ASSEMBLER_NAME_P too.
12829 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12830 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12831 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12832 * passes.c (rest_of_decl_compilation): Check
12833 HAS_DECL_ASSEMBLER_NAME_P too.
12834 * recog.c (verify_changes): Likewise.
12835 * tree-pretty-print.c (dump_decl_name): Likewise.
12836 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12838 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12839 (SET_DECL_ASSEMBLER_NAME): Use it.
12840 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12841 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12842 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
12844 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12846 * config.gcc (i386, x86_64): Add extra objects.
12847 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12848 (ix86_min_insn_size): Declare.
12849 (ix86_issue_rate): Declare.
12850 (ix86_adjust_cost): Declare.
12851 (ia32_multipass_dfa_lookahead): Declare.
12852 (ix86_macro_fusion_p): Declare.
12853 (ix86_macro_fusion_pair_p): Declare.
12854 (ix86_bd_has_dispatch): Declare.
12855 (ix86_bd_do_dispatch): Declare.
12856 (ix86_core2i7_init_hooks): Declare.
12857 (ix86_atom_sched_reorder): Declare.
12858 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12859 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12860 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12861 (rip_relative_addr_p): Rename to ...
12862 (ix86_rip_relative_addr_p): ... this one; export.
12863 (memory_address_length): Update.
12864 (ix86_issue_rate): Move to x86-tune-sched.c.
12865 (ix86_flags_dependent): Move to x86-tune-sched.c.
12866 (ix86_agi_dependent): Move to x86-tune-sched.c.
12867 (exact_dependency_1): Move to x86-tune-sched.c.
12868 (exact_store_load_dependency): Move to x86-tune-sched.c.
12869 (ix86_adjust_cost): Move to x86-tune-sched.c.
12870 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12871 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12872 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12873 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12874 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12875 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12876 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12877 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12878 (min_insn_size): Rename to ...
12879 (ix86_min_insn_size): ... this one; export.
12880 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12881 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12882 (core2i7_first_cycle_multipass_backtrack): Move to
12883 x86-tune-sched-core.c.
12884 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12885 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12886 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12887 (ix86_avoid_jump_mispredicts): Update.
12888 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12889 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12890 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12891 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12892 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12893 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12894 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12895 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12896 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12897 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12898 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12899 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12900 (BIG): Move to ix86-tune-sched-bd.c.
12901 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12902 (enum insn_path): Move to ix86-tune-sched-bd.c.
12903 (get_mem_group): Move to ix86-tune-sched-bd.c.
12904 (is_cmp): Move to ix86-tune-sched-bd.c.
12905 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12906 (is_branch): Move to ix86-tune-sched-bd.c.
12907 (is_prefetch): Move to ix86-tune-sched-bd.c.
12908 (init_window): Move to ix86-tune-sched-bd.c.
12909 (allocate_window): Move to ix86-tune-sched-bd.c.
12910 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12911 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
12912 (process_end_window): Move to ix86-tune-sched-bd.c.
12913 (allocate_next_window): Move to ix86-tune-sched-bd.c.
12914 (find_constant): Move to ix86-tune-sched-bd.c.
12915 (get_num_immediates): Move to ix86-tune-sched-bd.c.
12916 (has_immediate): Move to ix86-tune-sched-bd.c.
12917 (get_insn_path): Move to ix86-tune-sched-bd.c.
12918 (get_insn_group): Move to ix86-tune-sched-bd.c.
12919 (count_num_restricted): Move to ix86-tune-sched-bd.c.
12920 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
12921 (add_insn_window): Move to ix86-tune-sched-bd.c.
12922 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
12923 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
12924 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
12925 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
12926 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
12927 (do_dispatch): Move to ix86-tune-sched-bd.c.
12928 (has_dispatch): Move to ix86-tune-sched-bd.c.
12929 * config/i386/t-i386: Add new object files.
12930 * config/i386/x86-tune-costs.h: New file.
12931 * config/i386/x86-tune-sched-atom.c: New file.
12932 * config/i386/x86-tune-sched-bd.c: New file.
12933 * config/i386/x86-tune-sched-core.c: New file.
12934 * config/i386/x86-tune-sched.c: New file.
12936 2017-10-11 Liu Hao <lh_mouse@126.com>
12938 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
12939 the generic version below instead.
12940 (should_colorize): Recognize Windows consoles as terminals
12942 * pretty-print.c [__MINGW32__] (write_all): New function.
12943 [__MINGW32__] (find_esc_head): Likewise.
12944 [__MINGW32__] (find_esc_terminator): Likewise.
12945 [__MINGW32__] (eat_esc_sequence): Likewise.
12946 [__MINGW32__] (mingw_ansi_fputs): New function that handles
12948 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
12951 2017-10-11 Richard Biener <rguenther@suse.de>
12953 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
12954 Properly call analyze_scalar_evolution with the loop of the stmt.
12956 2017-10-11 Richard Biener <rguenther@suse.de>
12958 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
12959 * tree-core.h (tree_base): Add chrec_var union member.
12960 * tree.h (CHREC_VAR): Remove.
12961 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
12962 * tree-chrec.h (build_polynomial_chrec): Adjust.
12963 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
12964 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
12966 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
12968 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
12969 * match.pd: ... here.
12970 ((T) X == (T) Y): Relax condition.
12972 2017-10-11 Bin Cheng <bin.cheng@arm.com>
12974 PR tree-optimization/82472
12975 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
12977 (break_alias_scc_partitions): Update postorder number.
12979 2017-10-11 Martin Liska <mliska@suse.cz>
12982 * opts.c (parse_no_sanitize_attribute): Do not use error_value
12984 * opts.h (parse_no_sanitize_attribute): Remove last argument.
12986 2017-10-11 Martin Liska <mliska@suse.cz>
12988 * print-rtl.c (print_insn): Move declaration of idbuf
12989 to same scope as name.
12991 2017-10-11 Martin Liska <mliska@suse.cz>
12996 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
12997 volatile arguments.
12999 2017-10-11 Martin Liska <mliska@suse.cz>
13002 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
13003 volatile arguments.
13005 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
13007 * config.gcc (default_gnu_indirect_function): Default to yes for
13008 arm*-*-linux* with glibc.
13010 2017-10-11 Richard Biener <rguenther@suse.de>
13012 * tree-scalar-evolution.c (get_scalar_evolution): Handle
13013 default-defs and types we do not want to analyze.
13014 (interpret_loop_phi): Replace unreachable code with an assert.
13015 (compute_scalar_evolution_in_loop): Remove and inline ...
13016 (analyze_scalar_evolution_1): ... here, replacing condition with
13017 what makes the intent clearer. Remove handling of cases
13018 get_scalar_evolution now handles.
13020 2017-10-10 Jim Wilson <wilson@tuliptree.org>
13022 PR rtl-optimization/81434
13023 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
13024 comment for main loop. In sched_group_found if, also add checks for
13025 pass and min_cost_group.
13027 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
13029 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
13030 (rs6000_insn_cost): New function.
13031 * config/rs6000/rs6000.md (cost): New attribute.
13033 2017-10-10 Jakub Jelinek <jakub@redhat.com>
13034 H.J. Lu <hongjiu.lu@intel.com>
13038 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
13039 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
13040 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
13041 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
13042 to OPTION_MASK_ISA_AVX512VL - builtins that have both
13043 OPTION_MASK_ISA_MMX and some other bit set require both
13044 mmx and the ISAs without the mmx bit.
13045 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
13046 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
13047 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
13048 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
13049 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
13050 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
13051 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
13052 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
13053 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
13054 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
13055 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
13056 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
13057 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
13058 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
13059 Add OPTION_MASK_ISA_MMX.
13061 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
13063 * config.gcc (armv7*-*-freebsd*): New target.
13064 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
13066 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
13068 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
13069 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
13072 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
13074 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
13075 a template parameter.
13076 (WIDE_INT_REF_FOR): Update accordingly.
13077 * tree.h (wi::int_traits <const_tree>): Delete.
13078 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
13079 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
13080 (wi::tree_to_wide_ref): New typedef.
13081 (wi::to_wide): New function.
13082 * calls.c (get_size_range): Use wi::to_wide when operating on
13083 trees as wide_ints.
13084 * cgraph.c (cgraph_node::create_thunk): Likewise.
13085 * config/i386/i386.c (ix86_data_alignment): Likewise.
13086 (ix86_local_alignment): Likewise.
13087 * dbxout.c (stabstr_O): Likewise.
13088 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
13089 * expr.c (const_vector_from_tree): Likewise.
13090 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
13091 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
13092 (fold_negate_expr_1, int_const_binop_1, const_binop)
13093 (fold_convert_const_int_from_real, optimize_bit_field_compare)
13094 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
13095 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
13096 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
13097 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
13098 (fold_not_const, round_up_loc): Likewise.
13099 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
13100 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
13101 (alloca_call_type): Likewise.
13102 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
13103 * godump.c (go_output_typedef): Likewise.
13104 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
13105 * internal-fn.c (get_min_precision): Likewise.
13106 * ipa-cp.c (ipcp_store_vr_results): Likewise.
13107 * ipa-polymorphic-call.c
13108 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
13109 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
13110 (ipa_modify_call_arguments): Likewise.
13111 * match.pd: Likewise.
13112 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
13113 * print-tree.c (print_node_brief, print_node): Likewise.
13114 * stmt.c (expand_case): Likewise.
13115 * stor-layout.c (layout_type): Likewise.
13116 * tree-affine.c (tree_to_aff_combination): Likewise.
13117 * tree-cfg.c (group_case_labels_stmt): Likewise.
13118 * tree-data-ref.c (dr_analyze_indices): Likewise.
13119 (prune_runtime_alias_test_list): Likewise.
13120 * tree-dump.c (dequeue_and_dump): Likewise.
13121 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
13122 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
13123 * tree-pretty-print.c (dump_generic_node): Likewise.
13124 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
13125 (simple_iv_with_niters): Likewise.
13126 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
13127 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
13128 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
13129 * tree-ssa-loop-niter.c (split_to_var_and_offset)
13130 (refine_value_range_using_guard, number_of_iterations_ne_max)
13131 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
13132 (get_cst_init_from_scev, record_nonwrapping_iv)
13133 (scev_var_range_cant_overflow): Likewise.
13134 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
13135 * tree-ssa-pre.c (compute_avail): Likewise.
13136 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
13137 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
13138 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
13139 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
13140 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
13141 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
13142 (dump_case_nodes, try_switch_expansion): Likewise.
13143 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
13144 (vect_do_peeling): Likewise.
13145 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
13146 * tree-vect-stmts.c (vectorizable_load): Likewise.
13147 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
13148 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
13149 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
13150 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
13151 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
13152 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
13153 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
13154 (evrp_dom_walker::before_dom_children): Likewise.
13155 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
13156 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
13157 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
13158 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
13159 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
13160 (get_range_pos_neg): Likewise.
13161 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
13162 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
13163 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
13164 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
13165 * config/avr/avr.c (avr_fold_builtin): Likewise.
13166 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
13167 * config/msp430/msp430.c (msp430_attr): Likewise.
13168 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
13169 * config/powerpcspe/powerpcspe-c.c
13170 (altivec_resolve_overloaded_builtin): Likewise.
13171 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
13172 (rs6000_expand_ternop_builtin): Likewise.
13173 * config/rs6000/rs6000-c.c
13174 (altivec_resolve_overloaded_builtin): Likewise.
13175 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
13176 (rs6000_expand_ternop_builtin): Likewise.
13177 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
13179 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13181 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
13182 when copying loop nest with only one inner loop.
13184 2017-10-10 Richard Biener <rguenther@suse.de>
13186 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
13187 blocks if SCEV is active.
13188 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
13190 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
13191 (scev_initialize): Assert we are not yet initialized.
13193 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13195 * tree-loop-distribution.c (generate_loops_for_partition): Remove
13196 inner loop's exit stmt by making it always exit the loop, otherwise
13197 we would generate an infinite empty loop.
13199 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13201 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
13202 renaming variables in new preheader if it's deleted.
13204 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13206 * tree-loop-distribution.c (struct partition): Remove unused field
13207 loops of the structure.
13208 (partition_alloc, partition_free): Ditto.
13209 (build_rdg_partition_for_vertex): Ditto.
13211 2017-10-09 Jeff Law <law@redhat.com>
13213 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
13214 return type to match prototype and documentation.
13216 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13218 * config/rs6000/rs6000.c (processor_costs): Move to ...
13219 * config/rs6000/rs6000.h: ... here.
13220 (rs6000_cost): Declare.
13222 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
13224 * except.c (setjmp_fn): New global variable.
13225 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
13226 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
13227 if DONT_USE_BUILTIN_SETJMP is defined.
13229 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13231 * target.def (insn_cost): New hook.
13232 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
13233 * doc/tm.texi: Regenerate.
13234 * rtlanal.c (insn_cost): Use the new hook.
13236 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13238 * combine.c (combine_validate_cost): Compute the new insn_cost,
13239 not just pattern_cost.
13240 (try_combine): Adjust comment.
13242 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13244 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
13246 * combine.c (uid_insn_cost): Adjust comment.
13247 (combine_validate_cost): Adjust comment. Use pattern_cost instead
13249 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
13250 * dse.c (find_shift_sequence): Ditto.
13251 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
13252 (bb_valid_for_noce_process_p): Use pattern_cost.
13253 * rtl.h (insn_rtx_cost): Delete.
13254 (pattern_cost): New prototype.
13255 (insn_cost): New prototype.
13256 * rtlanal.c (insn_rtx_cost): Rename to...
13257 (pattern_cost): ... this.
13260 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
13262 * config/i386/i386.md (*jcc_2): Remove insn pattern.
13263 (*jcc<mode>_0_r_i387): Ditto.
13264 (*jccxf_r_i387): Ditto.
13265 (*jcc<mode>_r_i387): Ditto.
13266 (*jccu<mode>_r_i387): Ditto.
13267 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
13268 (*jcc): Rename from *jcc_1.
13270 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13272 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
13273 deferred rescans after the lvx/stvx recombination pre-pass.
13275 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
13277 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
13278 memory operation instruction support.
13279 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
13280 (rs6000-ibm-aix[789]*): Likewise.
13281 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
13282 Document new functions.
13284 2017-10-09 Richard Biener <rguenther@suse.de>
13286 PR tree-optimization/82397
13287 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
13288 equality only for semantically equal trees.
13290 2017-10-09 Richard Biener <rguenther@suse.de>
13292 PR tree-optimization/82449
13293 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
13294 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
13295 allow constant addresses.
13296 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
13299 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13301 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
13304 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13307 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
13310 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13313 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
13315 2017-10-09 Jakub Jelinek <jakub@redhat.com>
13318 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
13319 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
13321 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
13323 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
13324 (WI_BINARY_PREDICATE_RESULT): Likewise.
13325 (wi::binary_traits::operator_result): New type.
13326 (wi::binary_traits::predicate_result): Likewise.
13327 (generic_wide_int::operator~, unary generic_wide_int::operator-)
13328 (generic_wide_int::operator==, generic_wide_int::operator!=)
13329 (generic_wide_int::operator&, generic_wide_int::and_not)
13330 (generic_wide_int::operator|, generic_wide_int::or_not)
13331 (generic_wide_int::operator^, generic_wide_int::operator+
13332 (binary generic_wide_int::operator-, generic_wide_int::operator*):
13334 (operator~, unary operator-, operator==, operator!=, operator&)
13335 (operator|, operator^, operator+, binary operator-, operator*): New
13337 * expr.c (get_inner_reference): Use wi::bit_and_not.
13338 * fold-const.c (fold_binary_loc): Likewise.
13339 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
13340 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
13341 (bit_value_binop): Likewise.
13342 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
13343 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
13344 (extract_range_from_binary_expr_1): Likewise.
13345 (masked_increment): Likewise.
13346 (simplify_bit_ops_using_ranges): Likewise.
13348 2017-10-09 Martin Jambor <mjambor@suse.cz>
13351 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
13352 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
13353 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
13355 (hsa_fixup_mov_insn_type): New function.
13356 (hsa_op_with_type::get_in_type): Use it.
13357 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
13358 immediates in an assert.
13359 (hsa_op_with_type::extend_int_to_32bit): New method.
13360 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
13361 types. Convert to dest type if necessary.
13362 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
13363 (reg_for_gimple_ssa): Pass false as min32int to
13364 hsa_type_for_scalar_tree_type.
13365 (gen_hsa_addr): Fixup type when creating addresable temporary.
13366 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
13367 (gen_hsa_unary_operation): Extend operands and convert to dest type if
13368 necessary. Call hsa_fixup_mov_insn_type.
13369 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
13370 extend operands and convert to dest type if necessary.
13371 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
13372 to dest type if necessary.
13373 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
13374 if conversion nt necessary and size matches.
13375 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
13376 to dest type if necessary.
13377 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
13378 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
13380 (gen_hsa_clrsb): Likewise.
13381 (gen_hsa_ffs): Likewise.
13382 (gen_hsa_divmod): Extend operands and convert to dest type if
13384 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
13386 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13388 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
13389 Remove empty default arguments. Use a brace block as output
13391 (conditional return): Ditto.
13393 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
13394 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
13396 (group_ending_nop): Ditto.
13397 (doloop_end): Ditto.
13398 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
13399 (splitters for those): Ditto.
13401 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13403 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
13404 a conditional jump (and the compare for it) so that pc_rtx is the
13406 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
13407 for the deleted and renamed ctr<mode>_internal[234] patterns.
13408 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
13409 Delete second conditional return pattern.
13410 (ctr<mode>_internal2): Delete this second bdnz pattern.
13411 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
13412 (ctr<mode>_internal4): Delete this second bdz pattern.
13414 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13416 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
13417 (always_initialized_rtx_for_ssa_name_p): New predicate.
13418 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
13419 (finish_out_of_ssa): Free new field of SA.
13420 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
13421 * tree-ssa-coalesce.c: Include tree-ssa.h.
13422 (get_parm_default_def_partitions): Remove extern keyword.
13423 (get_undefined_value_partitions): New function.
13424 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
13425 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
13426 uninitialized bits.
13427 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
13429 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13431 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
13433 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13435 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
13436 for targets that preffer 128bit.
13438 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13440 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
13442 2017-10-08 Olivier Hainque <hainque@adacore.com>
13444 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
13445 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
13447 (thumb_set_return_address): Likewise.
13449 2017-10-08 Olivier Hainque <hainque@adacore.com>
13451 * common/config/arm/arm-common.c (arm_except_unwind_info):
13452 Handle DWARF2_UNWIND_INFO.
13454 2017-10-07 Michael Collison <michael.collison@arm.com>
13456 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
13459 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
13461 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
13462 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
13463 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
13464 defined, force the creation of a new block for a dispatch label.
13466 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
13468 * invoke.texi (Wsuggest-attribute=cold): Document.
13469 * common.opt (Wsuggest-attribute=cold): New
13470 * ipa-pure-const.c (warn_function_cold): New function.
13471 * predict.c (compute_function_frequency): Use it.
13472 * predict.h (warn_function_cold): Declare.
13474 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
13476 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
13479 2017-10-06 Martin Liska <mliska@suse.cz>
13481 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
13482 keyword for member functions.
13483 (struct sanopt_tree_couple): New struct.
13484 (struct sanopt_tree_couple_hash): New function.
13485 (struct sanopt_ctx): Add new hash_map.
13486 (has_dominating_ubsan_ptr_check): New function.
13487 (record_ubsan_ptr_check_stmt): Likewise.
13488 (maybe_optimize_ubsan_ptr_ifn): Likewise.
13489 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
13490 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
13492 2017-10-06 Sudakshina Das <sudi.das@arm.com>
13495 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
13496 aarch64_simd_valid_immediate on CONST_VECTORs.
13497 (aarch64_reg_or_bic_imm): Likewise.
13499 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
13501 PR rtl-optimization/82396
13502 * haifa-sched.c (ready_sort_real): Disable qsort checking.
13504 2017-10-06 Sebastian Pop <sebpop@gmail.com>
13506 * graphite-dependences.c (scop_get_reads): Move code to...
13507 (scop_get_must_writes): Move code to...
13508 (scop_get_may_writes): Move code to...
13509 (scop_get_reads_and_writes): ... here.
13510 (scop_get_dependences): Call scop_get_reads_and_writes.
13512 2017-10-06 Jakub Jelinek <jakub@redhat.com>
13514 PR tree-optimization/82434
13515 * fold-const.h (can_native_encode_type_p,
13516 can_native_encode_string_p): Remove.
13517 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
13518 don't encode anything, just return what would be otherwise returned.
13519 (native_encode_fixed, native_encode_complex, native_encode_vector):
13521 (native_encode_string): Likewise. Inline by hand
13522 can_native_encode_string_p.
13523 (can_native_encode_type_p): Remove.
13524 (can_native_encode_string_p): Remove.
13525 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
13526 STRING_CSTs using can_native_encode_string_p, test all
13527 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
13528 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
13529 argument from native_encode_expr.
13530 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
13531 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
13534 2017-10-06 Richard Biener <rguenther@suse.de>
13536 PR tree-optimization/82397
13537 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
13538 operand_equal_p but rely on data_ref_compare_tree for detecting
13540 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
13541 to match up with dr_group_sort_cmp.
13543 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13546 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
13548 * config/s390/s390-builtin-types.def: Regenerate.
13550 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13553 * config/s390/s390-builtin-types.def: Regenerate.
13554 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
13555 Change flag from B_VXE to B_VX.
13556 (s390_vec_min_dbl): Remove B_VXE flag.
13558 2017-10-06 Richard Biener <rguenther@suse.de>
13560 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
13561 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
13562 translate_isl_ast_to_gimple::is_valid_rename,
13563 translate_isl_ast_to_gimple::get_rename,
13564 translate_isl_ast_to_gimple::get_def_bb_for_const,
13565 translate_isl_ast_to_gimple::get_new_name,
13566 translate_isl_ast_to_gimple::collect_all_ssa_names,
13567 translate_isl_ast_to_gimple::copy_loop_phi_args,
13568 translate_isl_ast_to_gimple::collect_all_ssa_names,
13569 translate_isl_ast_to_gimple::copy_loop_phi_args,
13570 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
13571 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
13572 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
13573 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
13574 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
13575 translate_isl_ast_to_gimple::copy_cond_phi_args,
13576 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
13577 translate_isl_ast_to_gimple::edge_for_new_close_phis,
13578 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
13579 translate_isl_ast_to_gimple::rename_uses,
13580 translate_isl_ast_to_gimple::rename_all_uses): Remove.
13581 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
13582 (set_rename_for_each_def): Likewise.
13583 (graphite_copy_stmts_from_block): Handle debug stmt resetting
13584 here. Handle rewriting SCEV analyzable uses here.
13585 (copy_bb_and_scalar_dependences): Generate code for PHI
13587 (graphite_regenerate_ast_isl): Adjust.
13588 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
13589 (add_write, add_read): New functions.
13590 (build_cross_bb_scalars_def): Use it and simplify.
13591 (build_cross_bb_scalars_use): Likewise.
13592 (graphite_find_cross_bb_scalar_vars): Inline into...
13593 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
13594 simulating out-of-SSA. Compute liveout and add dependencies.
13595 (build_scops): Force an empty entry block.
13596 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
13598 (sese_build_liveouts): Declare.
13599 (sese_trivially_empty_bb_p): Likewise.
13600 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
13601 compute liveout and debug_liveout.
13602 (sese_bad_liveouts_use): Remove.
13603 (sese_reset_debug_liveouts_bb): Likewise.
13604 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
13605 (sese_build_liveouts): Build liveout and debug_liveout and store
13607 (new_sese_info): Adjust.
13608 (free_sese_info): Likewise.
13609 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
13610 do not build liveout here.
13611 (move_sese_in_condition): Adjust region entry.
13612 (scev_analyzable_p): Match up with chrec_apply requirements.
13613 (sese_trivially_empty_bb_p): New.
13614 * tree-into-ssa.c (get_reaching_def): Properly support generating
13615 default-defs for incremental rewrite of anonymous names.
13617 2017-10-06 Richard Biener <rguenther@suse.de>
13619 * graphite-sese-to-poly.c (extract_affine): For casts increasing
13620 precision do not perform modulo reduction.
13622 2017-10-06 Richard Biener <rguenther@suse.de>
13624 PR tree-optimization/82436
13625 * tree-vect-slp.c (vect_supported_load_permutation_p): More
13626 conservatively choose the vectorization factor when checking
13627 whether we can perform the required load permutation.
13628 (vect_transform_slp_perm_load): Assert when we may not fail.
13630 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
13632 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
13633 message for incompatible -msdata=* and -mcall-* options.
13635 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13637 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
13638 rate for post-reload scheduling.
13640 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13642 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
13644 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13646 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
13647 to improve monte carlo in scimark.
13649 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13651 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
13652 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
13653 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
13654 pentium4_cost, nocona_cost): Set reassociation width to 1.
13655 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
13656 width to 2 for fp operations and 1 otherwise.
13657 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
13659 (atom_cost): Set reassociation width to 2.
13660 (slm_cost, generic_cost): Set fp reassociation width
13661 to 2 and 1 otherwise.
13662 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
13663 (core_cost): Set fp reassociation width to 4 and vector to 2.
13664 (ix86_reassociation_width): Rewrite using cost table; special case
13665 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
13666 and TARGET_AVX128_OPTIMAL.
13667 * config/i386/i386.h (processor_costs): Add
13668 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
13669 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
13670 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
13671 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
13672 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
13673 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
13675 2017-10-05 Nathan Sidwell <nathan@acm.org>
13677 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
13679 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13681 * config/arm/arm.c (arm_test_fpu_data): New.
13682 (arm_run_selftests): Call arm_test_fpu_data.
13684 2017-10-04 Nathan Sidwell <nathan@acm.org>
13686 * toplev.c (toplev::main): Remove excess parens on pretty_printer
13688 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
13690 2017-10-04 Sudakshina Das <sudi.das@arm.com>
13692 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
13693 check type for aarch64_simd_valid_immediate.
13694 (aarch64_output_simd_mov_immediate): Update prototype.
13695 (aarch64_simd_valid_immediate): Update prototype.
13696 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
13697 support for ORR-immediate.
13698 (and<mode>3): modified pattern to add support for BIC-immediate.
13699 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
13700 now checks for valid immediate for BIC and ORR based on new enum
13702 (aarch64_output_simd_mov_immediate): Function now used to output
13703 BIC/ORR imm as well based on new enum argument.
13704 * config/aarch64/constraints.md (Do): New vector immediate constraint.
13706 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
13707 (aarch64_reg_or_bic_imm): Likewise.
13709 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13711 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
13712 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
13714 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13718 PR rtl-optimization/82396
13719 * haifa-sched.c (autopref_multipass_init): Simplify
13721 (autopref_rank_data): Simplify sort order.
13722 * sched-int.h (autopref_multipass_data_): Remove
13723 multi_mem_insn_p, min_offset and max_offset.
13725 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13727 * doc/sourcebuild.texi: Document vect_peeling_profitable.
13729 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13731 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
13732 vect_doubleint_cvt.
13734 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13736 * doc/sourcebuild.texi: Document vect_long_mult.
13738 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
13740 PR tree-optimization/82413
13741 * fold-const.c (build_range_check): Use widest_int when comparing
13742 the maximum ETYPE value with HIGH.
13744 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13746 PR rtl-optimization/82396
13747 * haifa-sched.c (autopref_multipass_init): Simplify
13749 (autopref_rank_data): Simplify sort order.
13750 * sched-int.h (autopref_multipass_data_): Remove
13751 multi_mem_insn_p, min_offset and max_offset.
13753 2017-10-04 Jakub Jelinek <jakub@redhat.com>
13755 PR tree-optimization/82381
13756 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
13757 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
13760 PR tree-optimization/82374
13761 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
13762 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
13763 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
13764 current_function_decl to the new decl.
13766 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
13768 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
13769 helper macro for IEEE float128 hardware built-in functions.
13770 (SQRTF128_ODD): Add built-in functions with the round-to-odd
13772 (TRUNCF128_ODD): Likewise.
13773 (ADDF128_ODD): Likewise.
13774 (SUBF128_ODD): Likewise.
13775 (MULF128_ODD): Likewise.
13776 (DIVF128_ODD): Likewise.
13777 (FMAF128_ODD): Likewise.
13778 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
13779 UNSPEC_TRUNC_ROUND_TO_ODD.
13780 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
13781 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
13782 floating point round to odd instructions.
13783 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
13784 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
13785 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
13786 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
13787 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
13788 (trunc<mode>sf2_hw): Change the truncate with round to odd
13789 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
13790 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
13791 to odd hardware instructions.
13792 (sub<mode>3_odd): Likewise.
13793 (mul<mode>3_odd): Likewise.
13794 (div<mode>3_odd): Likewise.
13795 (sqrt<mode>2_odd): Likewise.
13796 (fma<mode>4_odd): Likewise.
13797 (fms<mode>4_odd): Likewise.
13798 (nfma<mode>4_odd): Likewise.
13799 (nfms<mode>4_odd): Likewise.
13800 (trunc<mode>df2_odd): Change the truncate with round to odd
13801 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
13803 * doc/extend.texi (PowerPC built-in functions): Update documentation
13804 for existing IEEE float128-bit built-in functions. Add built-in
13805 functions that generate the IEEE 128-bit floating point round to
13808 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13810 PR rtl-optimization/77729
13811 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13812 to (X&(C1&~C2))|C2 transformations.
13814 2017-10-03 Martin Jambor <mjambor@suse.cz>
13816 PR tree-optimization/82363
13817 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13818 mismatch, mark lacc written regardless of racc.
13820 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13822 PR tree-optimization/82381
13823 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13824 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13825 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13826 fallthrough into reassoc_stmt_dominates_stmt_p.
13829 * combine.c (combine_instructions): Don't combine in unreachable
13832 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13835 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13836 function to not use the have_cpu variable. Do not set cpu_index,
13837 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13838 or the default cpu.
13839 (rs6000_valid_attribute_p): Remove duplicate initializations of
13840 old_optimize and func_optimize.
13841 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13842 (rs6000_activate_target_options): Make global.
13843 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
13846 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13848 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13849 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13850 Set *poffset to 0 and *psize and *pmax_size to -1 if
13851 *poffset + *psize overflows in HOST_WIDE_INT.
13853 PR tree-optimization/82387
13854 PR tree-optimization/82388
13855 PR tree-optimization/82389
13856 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13857 instead of live_bytes non-NULL.
13859 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13862 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13865 2017-10-02 Richard Biener <rguenther@suse.de>
13867 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13868 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13869 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13871 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13873 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13874 requested precision matches the type's.
13875 * calls.c (alloc_max_size): Calculate the new candidate size as
13876 a widest_int and use wi::to_widest when comparing it with the
13877 current candidate size.
13878 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13879 zero rather than integer_zero_node.
13880 * match.pd: Check for a no-op conversion before using wi::add
13881 rather than after. Use tree_to_uhwi when summing small shift
13882 counts into an unsigned int.
13884 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13885 Alan Hayward <alan.hayward@arm.com>
13886 David Sherwood <david.sherwood@arm.com>
13889 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13890 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13891 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13892 POINTER_AND_FP_REGS.
13894 2017-10-02 Richard Biener <rguenther@suse.de>
13896 PR tree-optimization/82355
13897 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13898 a mapping for the enclosing loop but avoid generating one for
13899 the loop tree root.
13900 (copy_bb_and_scalar_dependences): Remove premature codegen
13901 error on PHIs in blocks duplicated into multiple places.
13902 * graphite-scop-detection.c
13903 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13904 in the region use it as loop and nest to analyze the DR in.
13905 (try_generate_gimple_bb): Likewise.
13906 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13907 (add_loop_constraints): For blocks in a loop not in the region
13908 create a dimension with a single iteration.
13909 * sese.h (gbb_loop_at_index): Remove assert.
13911 2017-10-01 Kevin Buettner <kevinb@redhat.com>
13913 * omp-expand.c (adjust_context_scope): New function.
13914 (expand_parallel_call): Call adjust_context_scope.
13916 2017-10-01 Jeff Law <law@redhat.com>
13918 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
13919 dom_opt_dom_walker class with direct access to private members.
13920 Add comments. Call test_for_singularity.
13921 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
13922 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
13923 m_dummy_cond anymore.
13924 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
13926 (pass_dominator:execute): Build the dummy_cond here and pass it
13927 to the dom_opt_dom_walker ctor.
13928 (test_for_singularity): New function.
13930 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
13931 Maya Rashish <coypu@sdf.org>
13933 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
13934 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
13935 (alpha*-*-netbsd*) Use nbsd_tm_file.
13936 (arm*-*-netbsdelf*) Likewise.
13937 (i[34567]86-*-netbsdelf*) Likewise.
13938 (x86_64-*-netbsd*) Likewise.
13939 (mips*-*-netbsd*) Likewise.
13940 (powerpc-*-netbsd*) Likewise.
13941 (sh*-*-netbsd*) Likewise.
13942 (sparc-*-netbsdelf*) Likewise.
13943 (sparc64-*-netbsd*) Likewise.
13944 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
13946 (vax-*-netbsdelf*) Likewise.
13947 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
13948 (UINT_FAST8_TYPE) Likewise.
13949 (INT_FAST16_TYPE) Check CHAR_FAST16.
13950 (UINT_FAST16_TYPE) Likewise.
13952 2017-09-30 Jakub Jelinek <jakub@redhat.com>
13955 * config/i386/i386.md
13956 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
13957 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
13958 *divmodsi4_zext_2): New define_insn_and_split.
13959 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
13960 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
13961 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
13962 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
13963 New define_insn_and_split.
13964 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
13965 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
13966 operands[1] having DImode when mode is SImode.
13968 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
13969 always SImode for DIV and MOD in REG_EQUAL notes.
13971 2017-09-29 Yury Gribov <tetra2005@gmail.com>
13973 PR middle-end/82319
13974 * match.pd: Fix handling of NaNs in pattern.
13976 2017-09-29 Jeff Law <law@redhat.com>
13978 * sbitmap.c (bitmap_bit_in_range_p): New function.
13979 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
13980 * tree-ssa-dse.c (live_bytes_read): New function.
13981 (dse_classify_store): Ignore reads of dead bytes.
13983 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
13984 typos and whitespace errors.
13985 * config/i386/predicates.md (address_no_seg_operand): Likewise.
13986 * config/s390/s390.c (s390_emit_prologue): Likewise.
13988 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13991 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
13992 with a symbol for LRA.
13994 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
13996 PR rtl-optimization/82338
13997 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
13999 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
14001 * genmodes.c (calc_wider_mode): Suppress qsort macro.
14002 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
14003 (qsort_chk): Declare.
14004 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
14005 (qsort_chk): New function.
14007 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14009 PR tree-optimization/82337
14010 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
14011 phi definition if the PHI result appears in an abnormal PHI.
14012 (find_basis_for_base_expr): Don't record a basis if the LHS of the
14013 basis appears in an abnormal PHI.
14015 2017-09-29 Richard Biener <rguenther@suse.de>
14017 * graphite-isl-ast-to-gimple.c
14018 (translate_isl_ast_to_gimple::set_codegen_error): New function.
14019 (binary_op_to_tree): Use it.
14020 (get_rename_from_scev): Likewise.
14021 (copy_loop_phi_nodes): Likewise.
14022 (copy_bb_and_scalar_dependences): Likewise.
14023 (translate_pending_phi_nodes): Likewise.
14025 2017-09-29 Jakub Jelinek <jakub@redhat.com>
14028 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
14029 for movabsq $(i32 << shift), r64.
14031 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
14033 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
14034 index when encoding %esp as %rsp to avoid 0x67 prefix.
14036 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
14038 * config/i386/i386.md (*movsf_internal, *movdf_internal):
14039 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
14041 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
14043 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
14044 Extensions with more than 16 double VFP registers.
14045 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
14046 to_clear_mask and all code related to it. Replace the remaining
14047 entry by a sbitmap and adapt code accordingly.
14049 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
14051 * brig-builtins.def: Change pure attributes to const.
14053 2017-09-28 Joseph Myers <joseph@codesourcery.com>
14055 * config.gcc (default_gnu_indirect_function): Default to yes for
14056 sparc*-*-linux* with glibc.
14058 2017-09-28 Joseph Myers <joseph@codesourcery.com>
14060 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
14061 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
14062 when creating .init_array and .fini_array sections with priority
14065 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
14068 * config/aarch64/aarch64.c
14069 (aarch64_builtin_support_vector_misalignment): Always return false
14070 when misalignment is unknown.
14072 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
14074 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
14075 this function to return false if the definition used by the swap
14076 instruction is artificial, or if the memory address from which the
14077 constant value is loaded is not represented by a base address held
14078 in a register or if the base address register is a frame or stack
14079 pointer. Additionally, return false if the base address of the
14080 loaded constant is a SYMBOL_REF but is not considered to be a
14082 (replace_swapped_load_constant): New function.
14083 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
14084 loaded constant vector with a load of a swapped constant vector.
14086 2017-09-27 Carl Love <cel@us.ibm.com>
14088 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
14089 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
14090 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
14093 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
14095 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
14096 first, always call autopref_rank_data otherwise.
14098 2017-09-27 Richard Biener <rguenther@suse.de>
14100 * graphite-scop-detection.c (find_scop_parameters): Move
14101 loop bound handling ...
14102 (gather_bbs::before_dom_children): ... here, avoiding the need
14103 to build scop_info->loop_nest.
14104 (record_loop_in_sese): Remove.
14105 * sese.h (sese_info_t::loop_nest): Remove.
14106 * sese.c (new_sese_info): Do not allocate loop_nest.
14107 (free_sese_info): Do not free loop_nest.
14109 2017-09-27 Jakub Jelinek <jakub@redhat.com>
14112 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
14113 lhs from calls if the lhs has addressable type.
14115 2017-09-27 Richard Biener <rguenther@suse.de>
14117 * graphite.h (scop::max_alias_set): New member.
14118 * graphite-scop-detection.c: Remove references to non-existing
14119 --param in comments.
14120 (build_alias_sets): Record the maximum alias set used for drs.
14121 (build_scops): Support zero as unlimited for
14122 --param graphite-max-arrays-per-scop.
14123 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
14124 and inline into ...
14125 (build_poly_sr_1): ... here. Compute alias set based on the
14126 maximum alias set used for drs rather than
14127 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
14129 2017-09-27 Richard Biener <rguenther@suse.de>
14131 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
14132 --param loop-block-tile-size=0 to disable tiling.
14134 2017-09-27 Richard Biener <rguenther@suse.de>
14136 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
14137 (graphite-max-nb-scop-params): Document special value zero.
14138 * domwalk.h (dom_walker::STOP): New symbolical constant.
14139 (dom_walker::dom_walker): Add optional parameter for bb to
14141 (dom_walker::~dom_walker): Declare.
14142 (dom_walker::before_dom_children): Document STOP return value.
14143 (dom_walker::m_user_bb_to_rpo): New member.
14144 (dom_walker::m_bb_to_rpo): Likewise.
14145 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
14146 mapping here if not provided by the user.
14147 (dom_walker::~dom_walker): Free bb to RPO mapping if not
14148 provided by the user.
14149 (dom_walker::STOP): Define.
14150 (dom_walker::walk): Do not compute bb to RPO mapping here.
14151 Support STOP return value from before_dom_children to stop
14153 * graphite-optimize-isl.c (optimize_isl): If the schedule
14154 is the same still generate code if -fgraphite-identity
14155 or -floop-parallelize-all are given.
14156 * graphite-scop-detection.c: Include cfganal.h.
14157 (gather_bbs::gather_bbs): Get and pass through bb to RPO
14159 (gather_bbs::before_dom_children): Return STOP for BBs
14161 (build_scops): Compute bb to RPO mapping and pass it to
14162 the domwalk. Treat --param graphite-max-nb-scop-params=0
14163 as not limiting the number of params.
14164 * graphite.c (graphite_initialize): Remove limit on the
14165 number of basic-blocks in a function.
14166 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
14167 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
14168 default value of 10.
14170 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14172 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
14173 Adjust code to eliminate needing to do the shift right 32-bits
14174 operation after XSCVDPSPN.
14176 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14178 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
14179 ((X / Y) != 0 -> X >= Y): Likewise.
14181 2017-09-26 Carl Love <cel@us.ibm.com>
14183 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
14184 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
14185 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
14186 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
14187 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
14188 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
14189 definitions and overloading.
14190 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
14191 statement for P9V_BUILTIN_XST_LEN_R.
14192 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
14193 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
14194 define_expand and define_insn for the instructions and builtins.
14195 * doc/extend.texi: Update the built-in documentation file for the new
14196 built-in functions.
14197 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
14198 define_insn for the instructions
14200 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
14203 * gcc/config/netbsd-protos.h: New file.
14204 * gcc/config/netbsd.c: New file.
14205 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
14206 * gcc/config/t-netbsd: New file.
14207 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
14208 (tmake_file) Add t-netbsd.
14209 (extra_objs) Add netbsd.o.
14211 2017-09-26 Janus Weil <janus@gcc.gnu.org>
14215 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
14217 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14219 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
14220 sign extension from a vector register to a GPR by doing a 32-bit
14221 direct move and then an EXTSW.
14222 (extendsi<mode>2 splitter): Likewise.
14223 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
14224 right or vector extract after doing XSCVDPSPN. Use
14225 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
14227 (movdi_from_sf_zero_ext): Likewise.
14228 (reload_gpr_from_vsxsf): Likewise.
14229 (p8_mfvsrd_4_disf): Delete, no longer used.
14230 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
14231 then needing to move the SFmode to a GPR to use the XSCVDPSP
14232 instruction instead of FRSP and XSCVDPSPN.
14233 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
14234 it is adjacent to the other XSCVSPDP insns.
14235 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
14236 SFmode to be in traditional Altivec registers.
14237 (vsx_xscvdpspn): Eliminate useless alternative constraint.
14238 (vsx_xscvspdpn): Likewise.
14239 (vsx_xscvspdpn_scalar): Likewise.
14241 2017-09-26 Martin Jambor <mjambor@suse.cz>
14243 * tree-sra.c (compare_access_positions): Put integral types first,
14244 stabilize sorting of integral types, remove conditions putting
14245 non-full-precision integers last.
14246 (sort_and_splice_var_accesses): Disable scalarization if a
14247 non-integert would be represented by a non-full-precision integer.
14249 2017-09-26 Joseph Myers <joseph@codesourcery.com>
14251 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
14252 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
14253 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
14254 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
14255 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
14256 conditionals inside the function instead of around it. Call
14257 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
14258 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
14260 2017-09-26 Richard Biener <rguenther@suse.de>
14262 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
14264 (scop_detection::build_scop_breadth): ... this. Removed.
14265 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
14266 (scop_detection::harmful_stmt_in_bb): Likewise.
14267 (scop_detection::graphite_can_represent_stmt): Likewise.
14268 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
14269 (scop_detection::can_represent_loop): Remove recursion, fold in ...
14270 (scop_detection::can_represent_loop_1): ... this. Removed.
14271 (scop_detection::harmful_loop_in_region): Simplify after inlining
14272 the above and remove more quadraticness.
14273 (build_scops): Adjust.
14274 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
14277 2017-09-26 Jakub Jelinek <jakub@redhat.com>
14280 * config/i386/i386.c (ix86_print_operand_address_as): Only test
14281 REGNO (base) == SP_REG if base is a REG.
14283 PR middle-end/35691
14284 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
14285 if it is different SSA_NAME.
14286 (optimize_range_tests_cmp_bitwise): New function.
14287 (optimize_range_tests): Call it.
14289 2017-09-26 Richard Biener <rguenther@suse.de>
14291 PR tree-optimization/82321
14292 * graphite.c (canonicalize_loop_closed_ssa): Properly check
14293 for the def being inside the loop.
14295 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14297 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
14299 * config/s390/s390-builtins.def: Fix constraint on op4.
14301 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14303 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
14304 independent expanders.
14305 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
14306 ("vec_ordered", "vec_unordered"): New expanders.
14308 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14310 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
14313 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14315 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
14316 vec_unpacks_lo_v16qi.
14317 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
14319 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14321 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
14322 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
14323 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
14325 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14327 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
14329 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
14331 ("*vec_slb<mode>"): New insn pattern.
14332 ("vec_shr_<mode>"): New expander.
14333 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
14334 and force the shift count operand to V16QImode.
14335 ("vec_srb<mode>"): Set shift count mode to V16QI.
14337 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14339 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
14340 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
14341 ("vec_widen_smult_hi_<mode>"): New expander definitions.
14343 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
14346 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
14348 2017-09-26 Richard Biener <rguenther@suse.de>
14350 PR tree-optimization/82320
14351 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
14354 2017-09-25 Jeff Law <law@redhat.com>
14356 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
14357 prototype for new argument.
14358 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
14359 mostly extracted from rs6000_emit_allocate_stack.
14360 (rs6000_emit_probe_stack_range_stack_clash): New function.
14361 (rs6000_emit_allocate_stack): Call
14362 rs6000_emit_probe_stack_range_stack_clash as needed.
14363 (rs6000_emit_probe_stack_range): Add additional argument
14364 to call to gen_probe_stack_range{si,di}.
14365 (output_probe_stack_range): New.
14366 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
14367 (output_probe_stack_range_stack_clash): New.
14368 (rs6000_emit_prologue): Emit notes into dump file as requested.
14369 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
14370 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
14371 Add additional operand and pass it to output_probe_stack_range.
14373 2017-09-25 Bin Cheng <bin.cheng@arm.com>
14375 PR tree-optimization/82163
14376 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
14377 (checking_verify_loop_closed_ssa): New parameter.
14378 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
14379 (check_loop_closed_ssa_stmt): Delete.
14380 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
14381 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
14382 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
14385 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
14387 * brig-builtins.def: Treat HSAIL barrier builtins as
14388 setjmp/longjump style functions.
14390 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14392 * target.def (constant_alignment): New hook.
14393 * defaults.h (CONSTANT_ALIGNMENT): Delete.
14394 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
14395 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
14396 * doc/tm.texi: Regenerate.
14397 * targhooks.h (default_constant_alignment): Declare.
14398 (constant_alignment_word_strings): Likewise.
14399 * targhooks.c (default_constant_alignment): New function.
14400 (constant_alignment_word_strings): Likewise.
14401 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
14402 instead of CONSTANT_ALIGNMENT.
14403 * varasm.c (align_variable, get_variable_align, build_constant_desc)
14404 (force_const_mem): Likewise.
14405 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
14406 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
14407 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
14408 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14409 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
14411 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
14412 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14413 constant_alignment_word_strings.
14414 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
14415 (CONSTANT_ALIGNMENT): Likewise.
14416 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14417 (arm_constant_alignment): New function.
14418 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
14419 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14420 constant_alignment_word_strings.
14421 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
14422 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14423 constant_alignment_word_strings.
14424 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
14425 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14426 (cris_constant_alignment): New function.
14427 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
14428 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14429 (epiphany_constant_alignment): New function.
14430 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
14431 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14432 constant_alignment_word_strings.
14433 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
14434 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14435 constant_alignment_word_strings.
14436 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
14437 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14438 constant_alignment_word_strings.
14439 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
14440 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
14441 * config/i386/i386.c (ix86_constant_alignment): Make static.
14442 Use the same interface as the target hook.
14443 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14444 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
14445 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14446 constant_alignment_word_strings.
14447 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
14448 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
14449 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14450 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
14451 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14452 constant_alignment_word_strings.
14453 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
14454 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14455 constant_alignment_word_strings.
14456 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
14457 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14458 constant_alignment_word_strings.
14459 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
14460 * config/microblaze/microblaze.c (microblaze_constant_alignment):
14462 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14463 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
14464 * config/mips/mips.c (mips_constant_alignment): New function.
14465 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14466 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
14467 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
14468 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14469 (mmix_constant_alignment): Make static. Use the same interface
14470 as the target hook.
14471 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
14472 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14473 constant_alignment_word_strings.
14474 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
14475 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14476 constant_alignment_word_strings.
14477 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
14478 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14479 constant_alignment_word_strings.
14480 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
14481 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14482 (rs6000_constant_alignment): New function.
14483 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
14484 * config/riscv/riscv.c (riscv_constant_alignment): New function.
14485 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14486 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
14487 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14488 (rs6000_constant_alignment): New function.
14489 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
14490 * config/s390/s390.c (s390_constant_alignment): New function.
14491 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14492 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
14493 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14494 constant_alignment_word_strings.
14495 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
14496 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14497 (sparc_constant_alignment): New function.
14498 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
14499 * config/spu/spu.c (spu_constant_alignment): New function.
14500 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14501 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
14502 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14503 constant_alignment_word_strings.
14504 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
14505 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14506 constant_alignment_word_strings.
14507 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
14508 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14509 constant_alignment_word_strings.
14510 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
14511 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14512 (visium_constant_alignment): New function.
14513 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
14514 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14515 (xtensa_constant_alignment): New function.
14516 * system.h (CONSTANT_ALIGNMENT): Poison.
14518 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
14520 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14521 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
14522 (rs6000_builtin_valid_without_lhs): New helper function.
14523 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14524 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
14526 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14528 * target.h (vec_perm_indices): Use unsigned short rather than
14530 (auto_vec_perm_indices): Likewise.
14531 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
14532 Use unsigned int rather than unsigned char.
14533 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
14535 2017-09-25 Richard Biener <rguenther@suse.de>
14537 * cfgloop.h (sort_sibling_loops): Declare.
14538 * cfgloop.c (sort_sibling_loops_cmp): New helper.
14539 (sort_sibling_loops): New function sorting the sibling loop list
14541 * graphite.c (graphite_transform_loops): Sort sibling loops.
14543 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
14545 * target.def (vec_perm_const_ok): Change sel parameter to
14547 * optabs-query.c (can_vec_perm_p): Update accordingly.
14548 * doc/tm.texi: Regenerate.
14549 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
14550 auto_vec_perm_indices and remove separate nelt field.
14551 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
14552 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
14553 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
14554 (aarch64_expand_vec_perm_const): Update accordingly.
14555 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
14556 to vec_perm_indices.
14557 * config/arm/arm.c (expand_vec_perm_d): Change perm to
14558 auto_vec_perm_indices and remove separate nelt field.
14559 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
14560 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
14561 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
14563 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
14564 to vec_perm_indices.
14565 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
14566 sel to vec_perm_indices.
14567 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
14568 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
14569 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
14571 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
14574 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
14577 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
14578 on the FUNCTION_DECL function context if it has a DIE that is a
14581 2017-09-25 Richard Biener <rguenther@suse.de>
14583 PR tree-optimization/82285
14584 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
14587 2017-09-25 Tom de Vries <tom@codesourcery.com>
14591 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
14594 2017-09-25 Richard Biener <rguenther@suse.de>
14596 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
14597 ISL errors other than isl_error_quota happen. Dump if the
14598 schedule is the same.
14599 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
14600 errors instead of aborting inside ISL.
14602 2017-09-25 Iain Sandoe <iain@codesourcery.com>
14605 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
14606 of libgcc_eh for m64.
14607 * config/i386/darwin64.h: Likewise.
14609 2017-09-25 Richard Biener <rguenther@suse.de>
14611 PR middle-end/82144
14612 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
14613 attribute for incomplete types nor twice for complete ones.
14615 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
14618 * config/i386/i386.c (ix86_print_operand_address_as): Encode
14619 %esp as %rsp to avoid 0x67 prefix if there is no index or base
14622 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
14625 * config/i386/i386.opt (mprefer-avx256): Use
14626 ix86_target_flags variable.
14627 * config/i386/i386.c (ix86_target_string): Move
14628 -mprefer-avx256 to flag2_opts.
14630 2017-09-22 Jakub Jelinek <jakub@redhat.com>
14632 PR middle-end/35691
14633 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
14634 and x != -1 | y != -1 into (x & y) != -1.
14636 2017-09-22 Steve Ellcey <sellcey@cavium.com>
14638 * config.gcc: Add new case statement to set
14639 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
14640 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
14641 s390x-*-linux* case statements. Added aarch64 to the list of
14642 supported architectures.
14644 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14646 PR tree-optimization/82289
14647 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
14648 STMT_VINFO_RELEVANT_P.
14650 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14651 Alan Hayward <alan.hayward@arm.com>
14652 David Sherwood <david.sherwood@arm.com>
14654 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
14655 for VR_RANGE only; don't allow VR_ANTI_RANGE.
14656 (extract_range_from_binary_expr_1): Don't call
14657 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
14659 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14660 Alan Hayward <alan.hayward@arm.com>
14661 David Sherwood <david.sherwood@arm.com>
14663 * target.def (preferred_vector_alignment): New hook.
14664 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
14666 * doc/tm.texi: Regenerate.
14667 * targhooks.h (default_preferred_vector_alignment): Declare.
14668 * targhooks.c (default_preferred_vector_alignment): New function.
14669 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
14671 (DR_TARGET_ALIGNMENT): New macro.
14672 (aligned_access_p): Update commentary.
14673 (vect_known_alignment_in_bytes): New function.
14674 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
14676 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
14677 Calculate the misalignment based on the target alignment rather than
14679 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
14680 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
14681 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
14682 the target alignment, rather than masking the element misalignment
14683 with the number of elements in a vector. Also use the target
14684 alignment when calculating the maximum number of peels.
14685 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
14686 instead of TYPE_ALIGN_UNIT.
14687 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
14688 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
14689 (vect_create_addr_base_for_vector_ref): Update call accordingly.
14690 (vect_create_data_ref_ptr): Likewise.
14691 (vect_setup_realignment): Realign by ANDing with
14692 -DR_TARGET_MISALIGNMENT.
14693 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
14694 the number of peels based on DR_TARGET_ALIGNMENT.
14695 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
14696 with the guaranteed alignment boundary when deciding whether
14698 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
14699 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
14700 (ensure_base_align): Remove stmt_info parameter. Get the
14701 target base alignment from DR_TARGET_ALIGNMENT.
14702 (vectorizable_store): Update call accordingly. Interpret
14703 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
14705 (vectorizable_load): Likewise.
14707 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14708 Alan Hayward <alan.hayward@arm.com>
14709 David Sherwood <david.sherwood@arm.com>
14711 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
14712 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
14713 (vect_enhance_data_refs_alignment): Likewise.
14715 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14717 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14718 error. Only quit immediately if parsing is complete.
14719 (BEGIN): Initialize fatal_err and parse_done.
14720 (begin fpu, end fpu): Check number of arguments.
14721 (begin arch, end arch): Likewise.
14722 (begin cpu, end cpu): Likewise.
14723 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14724 (optalias): Likewise.
14726 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14728 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14729 * config/arm/arm-isa.h: Delete. Move definitions to ...
14730 * arm-cpus.in: ... here. Use new feature and fgroup values.
14731 * config/arm/arm.c (arm_option_override): Use lower case for feature
14733 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14734 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14735 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14737 (print_isa_bits_for): New function.
14738 (gen_isa): New function.
14739 (gen_comm_data): Use print_isa_bits_for.
14740 (define feature): New keyword.
14741 (define fgroup): New keyword.
14742 * config/arm/t-arm (TM_H): Remove.
14743 (GTM_H): Add arm-isa.h.
14744 (arm-isa.h): Add rule to generate file.
14745 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14746 case for feature bit names.
14748 2017-09-22 Richard Biener <rguenther@suse.de>
14750 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
14752 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
14753 print of no dependency loops ...
14754 * graphite.c (graphite_transform_loops): ... here.
14755 (canonicalize_loop_closed_ssa_form): Work from inner to outer
14757 (same_close_phi_node, remove_duplicate_close_phi,
14758 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
14759 (canonicalize_loop_closed_ssa): ... here and simplify.
14760 * graphite-optimize-isl.c: Include tree-vectorizer.h.
14761 (optimize_isl): Use dump_printf_loc to tell when we stopped
14762 optimizing because of an ISL timeout.
14764 2017-09-22 Richard Biener <rguenther@suse.de>
14766 PR tree-optimization/82291
14767 * tree-if-conv.c (predicate_mem_writes): Make sure to
14768 remove writes in blocks predicated with false.
14770 2017-09-22 Richard Biener <rguenther@suse.de>
14772 * sese.c: Include cfganal.h.
14773 (if_region_set_false_region): Remove.
14774 (create_if_region_on_edge): Likewise.
14775 (move_sese_in_condition): Re-implement without destroying
14778 2017-09-22 Richard Biener <rguenther@suse.de>
14780 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
14781 Verify both BBs contain loop PHI nodes before dispatching to
14782 copy_loop_phi_args.
14783 (graphite_regenerate_ast_isl): Do not recompute dominators,
14784 do not verify three times. Restructure for clarity.
14785 * graphite-scop-detection.c (same_close_phi_node,
14786 remove_duplicate_close_phi, make_close_phi_nodes_unique,
14787 defined_in_loop_p, canonicalize_loop_closed_ssa,
14788 canonicalize_loop_closed_ssa_form): Simplify, remove excess
14789 checking and SSA rewrite, move to ...
14790 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
14791 (graphite_initialize): Do not pass in ctx, do not reset the
14792 SCEV cache, compute only dominators.
14793 (graphite_transform_loops): Allocate ISL ctx after
14794 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
14795 Maintain post-dominators only around build_scops.
14796 * sese.c (if_region_set_false_region): Make static. Free
14797 and recompute dominators.
14798 (move_sese_in_condition): Assert we don't get called with
14799 post-dominators computed.
14800 * sese.h (if_region_set_false_region): Remove.
14802 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
14804 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
14805 mode attribute for TARGET_AVX512VL.
14807 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14809 * config/i386/i386.opt (mprefer-avx256): New option.
14810 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14812 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14813 for TARGET_PREFER_AVX256.
14814 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14816 2017-09-21 Jeff Law <law@redhat.com>
14818 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14819 Fix dump output if the only stack space is for pushed registers.
14821 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14823 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14826 2017-09-21 Martin Sebor <msebor@redhat.com>
14829 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14830 code (in C++) or code that triggers warnings.
14832 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14834 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14836 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14838 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14839 * sched-rgn.c: Ditto.
14840 * sel-sched-ir.c: Ditto.
14842 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
14844 * toplev.h (set_random_seed): Adjust return type.
14845 * toplev.c (init_local_tick): Move eager initialization of random_seed
14846 to get_random_seed. Adjust comment.
14847 (init_random_seed): Inline to get_random_seed, delete.
14848 (get_random_seed): Initialize random_seed lazily.
14849 (set_random_seed): Do not return previous value.
14850 (print_switch_value): Do not call get_random_seed.
14852 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14854 * cgraph.c (delete_function_version): New, broken out from...
14855 (cgraph_node::delete_function_version): ...here. Rename to
14856 cgraph_node::delete_function_version_by_decl. Update all uses.
14857 (cgraph_node::remove): Call delete_function_version.
14859 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14862 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14863 VAR_DECLs to which addressable non-volatile parameters are mapped
14864 and for id->retvar after the return value assignment. Clear
14865 id->retval and id->retbnd after inlining.
14867 2017-09-21 Richard Biener <rguenther@suse.de>
14869 PR tree-optimization/82276
14870 PR tree-optimization/82244
14871 * tree-vrp.c (build_assert_expr_for): Set
14872 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14874 (remove_range_assertions): Revert earlier change.
14876 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14879 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14881 2017-09-21 Richard Biener <rguenther@suse.de>
14883 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14884 Restore valid IL after code generation errors.
14885 * graphite.c (graphite_transform_loops): Diagnose code
14886 generation issues as MSG_MISSED_OPTIMIZATION and continue
14887 with processing SCOPs.
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 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14894 rather than choosing between plus_constant and gen_rtx_<CODE>.
14895 * expr.c (emit_push_insn): Likewise.
14896 (expand_expr_real_2): Likewise.
14898 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14899 Alan Hayward <alan.hayward@arm.com>
14900 David Sherwood <david.sherwood@arm.com>
14902 * loop-unroll.c (split_iv): Call copy_rtx on the step.
14904 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14905 Alan Hayward <alan.hayward@arm.com>
14906 David Sherwood <david.sherwood@arm.com>
14908 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14909 calling tree_to_uhwi.
14911 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14912 Alan Hayward <alan.hayward@arm.com>
14913 David Sherwood <david.sherwood@arm.com>
14915 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
14916 INTEGER_CST rather than a negative test for ADDR_EXPR.
14918 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14919 Alan Hayward <alan.hayward@arm.com>
14920 David Sherwood <david.sherwood@arm.com>
14922 * tree-vrp.c (extract_range_from_binary_expr_1): Check
14923 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
14925 2017-09-21 Richard Biener <rguenther@suse.de>
14927 PR tree-optimization/71351
14928 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
14929 graphite_create_new_loop_guard): Remove, fold remaining parts
14931 (translate_isl_ast_node_for): ... here and simplify.
14933 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14936 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
14937 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
14938 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
14939 alternative always use QI mode, for -Os imov (=R,R) alternative
14940 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
14943 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14944 Jeff Law <law@redhat.com>
14946 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
14947 (allocate_stack_space): New function, partially extracted from
14948 s390_emit_prologue.
14949 (s390_emit_prologue): Track offset to most recent stack probe.
14950 Code to allocate space moved into allocate_stack_space.
14951 Dump actions when no stack is allocated.
14952 (s390_prologue_plus_offset): New function.
14953 (s390_emit_stack_probe): Likewise.
14955 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
14957 * common.opt (Wa, Wl, Wp, g, gz=): Add
14959 (gno-column-info): Remove.
14960 (gcolumn-info): Drop RejectNegative.
14961 (gno-): New prefix.
14962 (gno-record-gcc-switches): Remove.
14963 (grecord-gcc-switches): Drop RejectNegative.
14964 (gno-split-dwarf): Remove.
14965 (gsplit-dwarf): Drop RejectNegative.
14966 (gno-strict-dwarf): Remove.
14967 (gstrict-dwarf): Drop RejectNegative.
14968 * config/darwin.opt (gfull, gused): Add RejectNegative.
14969 * dwarf2out.c (gen_producer_string): Drop
14970 gno-record-gcc-switches handler.
14971 * optc-gen.awk: Add g to prefixes with negative forms.
14972 * opts-common.c (remapping_prefix_p): New.
14973 (find_opt): Check it.
14974 (generate_canonical_option): Test g prefix.
14975 (option_map): Add -gno- mapping.
14976 (add_misspelling_candidates): Check remapping_prefix_p.
14978 2017-09-20 Jeff Law <law@redhat.com>
14980 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
14981 thinko in stack clash protection support.
14983 * explow.c (compute_stack_clash_protection_loop_data): Use
14984 CONST_INT_P instead of explicit test. Verify object is a
14985 CONST_INT_P before looking at INTVAL.
14986 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
14987 instead of explicit test.
14989 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
14992 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
14993 address instead of to r1 and r11.
14995 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
14997 * config.gcc: Support "knm".
14998 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
14999 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
15001 * config/i386/i386.c (m_KNM): Define.
15002 (processor_target_table): Add "knm".
15004 (ix86_option_override_internal): Add "knm".
15005 (ix86_issue_rate): Add PROCESSOR_KNM.
15006 (ix86_adjust_cost): Ditto.
15007 (ia32_multipass_dfa_lookahead): Ditto.
15008 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
15009 (fold_builtin_cpu): Add M_INTEL_KNM.
15010 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
15011 (processor_type): Add PROCESSOR_KNM.
15012 * config/i386/x86-tune.def: Add m_KNM.
15013 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
15015 2017-09-20 Richard Biener <rguenther@suse.de>
15017 PR tree-optimization/80213
15018 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
15019 are allowed in empty BBs as well.
15020 (canonicalize_loop_closed_ssa): Also look for other complex
15022 (scop_detection::get_sese): Include the loop-closed PHI block
15024 (scop_detection::merge_sese): Remove code adding extra blocks.
15025 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
15026 (build_scops): Assert the final returned scop is invalid.
15028 2017-09-20 Richard Biener <rguenther@suse.de>
15030 PR tree-optimization/82264
15031 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
15033 (vn_phi_lookup): Likewise.
15034 (vn_phi_insert): Likewise.
15036 2017-09-20 Jakub Jelinek <jakub@redhat.com>
15038 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
15039 that fits into uhwi or shwi, add DW_AT_const_value regardless
15040 of early_dwarf without going through RTL, using add_AT_unsigned
15043 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
15044 (DEBUG_LTO_ABBREV_SECTION): Likewise.
15045 (DEBUG_LTO_MACINFO_SECTION): Likewise.
15046 (DEBUG_MACRO_SECTION): Likewise.
15047 (DEBUG_LTO_MACRO_SECTION): Likewise.
15048 (DEBUG_STR_DWO_SECTION): Likewise.
15049 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
15050 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
15051 (DEBUG_LTO_DWO_LINE_SECTION): Define.
15052 (DEBUG_LTO_LINE_STR_SECTION): Define.
15053 (init_sections_and_labels): Initialize debug_line_str_section
15054 variable. Initialize debug_loc_section for -gdwarf-5 to
15055 DEBUG_LOCLISTS_SECTION. Formatting fixes.
15057 2017-09-20 Richard Biener <rguenther@suse.de>
15059 * graphite-sese-to-poly.c (extract_affine): Properly handle
15060 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
15062 2017-09-20 Richard Biener <rguenther@suse.de>
15064 PR tree-optimization/81373
15065 * graphite-scop-detection.c (build_cross_bb_scalars_def):
15066 Force SESE live-out defs to be handled even if they are
15069 2017-09-19 Jeff Law <law@redhat.com>
15071 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
15072 nothing for stack adjustments with REG_STACK_CHECK.
15073 * sched-deps.c (parse_add_or_inc): Reject insns with
15074 REG_STACK_CHECK from dependency breaking.
15075 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
15076 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
15077 * reg-notes.def (STACK_CHECK): New note.
15079 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
15080 (ix86_expand_prologue): Dump stack clash info as needed.
15081 Call ix86_adjust_stack_and_probe_stack_clash as needed.
15083 * function.c (dump_stack_clash_frame_info): New function.
15084 * function.h (dump_stack_clash_frame_info): Prototype.
15085 (enum stack_clash_probes): New enum.
15087 * config/alpha/alpha.c (alpha_expand_prologue): Also check
15088 flag_stack_clash_protection.
15089 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
15090 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
15091 (arm_frame_pointer_required): Likewise.
15092 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
15093 (ia64_expand_prologue): Likewise.
15094 * config/mips/mips.c (mips_expand_prologue): Likewise.
15095 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
15096 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15097 (sparc_flat_expand_prologue): Likewise.
15098 * config/spu/spu.c (spu_expand_prologue): Likewise.
15100 * explow.c: Include "params.h".
15101 (anti_adjust_stack_and_probe_stack_clash): New function.
15102 (get_stack_check_protect): Likewise.
15103 (compute_stack_clash_protection_loop_data): Likewise.
15104 (emit_stack_clash_protection_loop_start): Likewise.
15105 (emit_stack_clash_protection_loop_end): Likewise.
15106 (allocate_dynamic_stack_space): Use get_stack_check_protect.
15107 Use anti_adjust_stack_and_probe_stack_clash.
15108 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
15109 (emit_stack_clash_protection_loop_start): Likewise.
15110 (emit_stack_clash_protection_loop_end): Likewise.
15111 * rtl.h (get_stack_check_protect): Prototype.
15112 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
15113 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
15114 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
15116 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
15118 * doc/tm.texi: Rebuilt.
15119 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
15120 get_stack_check_protect.
15121 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
15122 * config/arm/arm.c (arm_expand_prologue): Likewise.
15123 (arm_frame_pointer_required): Likewise.
15124 * config/i386/i386.c (ix86_expand_prologue): Likewise.
15125 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
15126 * config/mips/mips.c (mips_expand_prologue): Likewise.
15127 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
15128 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
15129 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15130 (sparc_flat_expand_prologue): Likewise.
15132 * common.opt (-fstack-clash-protection): New option.
15133 * flag-types.h (enum stack_check_type): Note difference between
15134 -fstack-check= and -fstack-clash-protection.
15135 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
15136 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
15137 * toplev.c (process_options): Issue warnings/errors for cases
15138 not handled with -fstack-clash-protection.
15139 * doc/invoke.texi (-fstack-clash-protection): Document new option.
15140 (-fstack-check): Note additional problem with -fstack-check=generic.
15141 Note that -fstack-check is primarily for Ada and refer users
15142 to -fstack-clash-protection for stack-clash-protection.
15143 Document new params for stack clash protection.
15145 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
15147 * config/i386/i386.c (ix86_split_long_move): Do not handle
15148 address used for LEA in a special way.
15150 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
15152 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
15154 2017-09-19 Martin Sebor <msebor@redhat.com>
15157 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
15158 of incompatible types.
15160 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
15162 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15163 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
15164 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15165 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
15167 2017-09-19 Richard Biener <rguenther@suse.de>
15169 PR tree-optimization/82244
15170 * tree-vrp.c (remove_range_assertions): Do not propagate
15171 a constant to abnormals but replace the assert with a copy.
15173 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
15175 PR rtl-optimization/57878
15176 PR rtl-optimization/68988
15177 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
15178 avoidance test involving non_reload_pseudos. Move frequency test
15179 below the general fragmentation avoidance test.
15181 2017-09-19 Richard Biener <rguenther@suse.de>
15183 PR tree-optimization/69728
15184 * graphite-sese-to-poly.c (schedule_error): New global.
15185 (add_loop_schedule): Handle empty domain by failing the
15187 (build_original_schedule): Handle schedule_error.
15189 2017-09-19 Richard Biener <rguenther@suse.de>
15191 * graphite-scop-detection.c (scop_detection::can_represent_loop):
15192 Do not iterate to sibling loops but only to siblings of inner
15195 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
15198 * config/m68k/m68k.md (moveq feeding equality comparison): Check
15199 that the registers are different.
15201 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
15203 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
15204 to processor_model and "amdfam17h" to arch_names_table.
15205 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
15207 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15210 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
15212 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15213 Alan Hayward <alan.hayward@arm.com>
15214 David Sherwood <david.sherwood@arm.com>
15216 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
15218 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
15220 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
15221 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
15222 vect_schedule_slp_instance.
15223 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
15224 Update call to vect_slp_analyze_node_operations. Simplify return
15226 (vect_slp_analyze_bb_1): Update call accordingly.
15227 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
15228 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
15229 (vect_schedule_slp): Update call accordingly.
15231 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15232 Alan Hayward <alan.hayward@arm.com>
15233 David Sherwood <david.sherwood@arm.com>
15235 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
15236 with types that aren't in fact scalar.
15238 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15240 * tree-vect-slp.c (vect_record_max_nunits): New function,
15242 (vect_build_slp_tree_1): ...here.
15243 (vect_build_slp_tree_2): Call it for phis too.
15245 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15247 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
15248 to vect_get_vec_def_for_operand when getting the mask operand.
15250 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15251 Alan Hayward <alan.hayward@arm.com>
15252 David Sherwood <david.sherwood@arm.com>
15254 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
15257 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15258 Alan Hayward <alan.hayward@arm.com>
15259 David Sherwood <david.sherwood@arm.com>
15261 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
15262 calculation for vector booleans.
15264 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15265 Alan Hayward <alan.hayward@arm.com>
15266 David Sherwood <david.sherwood@arm.com>
15268 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
15270 (vect_transform_stmt): ...here.
15271 (vect_analyze_stmt): Use it instead of calling
15272 vectorizable_live_operation directly.
15274 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
15276 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
15277 non-SIMT targets in acc vector loops.
15279 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
15281 * configure.ac: Add arc and check if assembler supports gdwarf2.
15282 * configure: Regenerate.
15284 2017-09-18 Richard Biener <rguenther@suse.de>
15286 PR tree-optimization/82220
15287 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
15288 epilogue niters from the min_profitable_iters compute.
15290 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15293 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
15294 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
15295 (ix86_init_pic_reg): Revert 2017-09-01 changes.
15297 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
15300 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
15301 switching to a new text section.
15303 2017-09-18 Richard Biener <rguenther@suse.de>
15305 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
15307 (build_alias_set): Reject aliases with no access function.
15309 2017-09-18 Richard Biener <rguenther@suse.de>
15311 PR tree-optimization/79622
15312 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
15314 (build_cross_bb_scalars_use): Likewise.
15316 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
15318 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
15320 2017-09-18 Alan Modra <amodra@gmail.com>
15323 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
15324 stack_pointer_rtx for count 0. Update comments. Break up
15325 large rtl expression.
15327 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
15329 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
15330 Increase to 20 bytes.
15331 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
15332 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
15333 or avx version of the stub.
15335 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
15338 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
15339 compute the minimum stack alignment. Also update preferred stack
15340 boundary for leaf functions.
15342 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
15344 PR tree-optimization/82228
15345 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
15348 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15350 * common/config/nds32/nds32-common.c
15351 (nds32_option_optimization_table): Refine formatting.
15352 (nds32_option_optimization_table): Use -fsched-pressure and
15353 -fomit-frame-pointer for specific optimization level.
15355 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15357 * config/nds32/nds32.c: Refine formatting and comments.
15358 * config/nds32/nds32.h: Likewise.
15359 * config/nds32/nds32.md: Likewise.
15360 * config/nds32/nds32-cost.c: Likewise.
15361 * config/nds32/nds32-isr.c: Likewise.
15362 * config/nds32/nds32-md-auxiliary.c: Likewise.
15363 * config/nds32/nds32-multiple.md: Likewise.
15364 * config/nds32/nds32-predicates.c: Likewise.
15366 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
15367 Jakub Jelinek <jakub@redhat.com>
15369 Add support for -std=c++2a.
15370 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
15372 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
15374 2017-09-15 Steve Ellcey <sellcey@cavium.com>
15377 * doc/extend.texi (Common Function Attributes): Add
15378 references to ARM, AArch64, and S/390 specific attributes.
15379 (Function Specific Option Pragmas): Add AArch64 and S/390
15380 to list of back ends that support the target pragma.
15382 2017-09-15 Nathan Sidwell <nathan@acm.org>
15384 * doc/standards.texi: Fix C++17 description. Update URLs for
15387 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
15389 * common.opt (Wcast-align=strict): New warning option.
15390 * doc/invoke.texi: Document -Wcast-align=strict.
15392 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
15394 * cgraph.h (cgraph_thunk_info): Add comments.
15395 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
15396 assert for VIRTUAL_* arguments stricter.
15398 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
15400 PR tree-optimization/71026
15401 * match.pd: Move RDIV patterns from fold-const.c
15402 * fold-const.c (distribute_real_division): Removed.
15403 (fold_binary_loc): Remove calls to distribute_real_divison.
15405 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15407 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
15408 c++1z and gnu++1z as deprecated. Change other references to
15409 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
15410 Change -Wc++1z-compat to -Wc++17-compat.
15411 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
15412 * dwarf2out.c (highest_c_language): Handle C++17.
15413 (gen_compile_unit_die): Likewise.
15415 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15417 PR rtl-optimization/82192
15418 * combine.c (make_extraction): Don't look through non-paradoxical
15419 SUBREGs or TRUNCATE if pos + len is or might be bigger than
15422 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15423 Alan Hayward <alan.hayward@arm.com>
15424 David Sherwood <david.sherwood@arm.com>
15426 * target.def (function_arg_offset): New hook.
15427 * targhooks.h (default_function_arg_offset): Declare.
15428 * targhooks.c (default_function_arg_offset): New function.
15429 * function.c (locate_and_pad_parm): Use
15430 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
15431 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
15432 (TARGET_FUNCTION_ARG_OFFSET): ...this.
15433 * doc/tm.texi: Regenerate.
15434 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
15435 * config/spu/spu.c (spu_function_arg_offset): New function.
15436 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
15437 * system.h (FUNCTION_ARG_OFFSET): Poison.
15439 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15440 Alan Hayard <alan.hayward@arm.com>
15441 David Sherwood <david.sherwood@arm.com>
15443 * target.def (truly_noop_truncation): New hook.
15444 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
15445 than TRULY_NOOP_TRUNCATION.
15446 * hooks.h (hook_bool_uint_uint_true): Declare.
15447 * hooks.c (hook_bool_uint_uint_true): New function.
15448 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
15449 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
15450 * doc/tm.texi: Regenerate.
15451 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
15452 rather than TRULY_NOOP_TRUNCATION in comments.
15453 (simplify_comparison): Likewise.
15454 (record_truncated_value): Likewise.
15455 * expmed.c (extract_bit_field_1): Likewise.
15456 (extract_split_bit_field): Likewise.
15457 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
15458 instead of TRULY_NOOP_TRUNCATION.
15459 * function.c (assign_parm_setup_block): Likewise.
15460 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
15461 * rtlhooks.c: Include target.h.
15462 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
15463 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
15464 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
15465 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
15466 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
15467 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
15468 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
15469 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
15470 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
15471 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
15472 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
15473 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
15474 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
15475 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
15476 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
15477 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
15478 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
15479 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
15480 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
15481 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
15482 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
15483 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
15484 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
15485 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
15486 * config/mips/mips.c (mips_truly_noop_truncation): New function.
15487 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15488 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15489 rather than TRULY_NOOP_TRUNCATION in comments.
15490 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
15491 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
15492 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
15493 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
15494 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
15495 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
15496 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
15497 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
15498 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
15499 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
15500 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
15501 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15502 rather than TRULY_NOOP_TRUNCATION in comments.
15503 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
15504 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
15505 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
15506 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
15507 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
15508 TRULY_NOOP_TRUNCATION condition.
15509 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
15510 (TRULY_NOOP_TRUNCATION): Delete.
15511 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
15512 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
15513 * config/spu/spu.c (spu_truly_noop_truncation): New function.
15514 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15515 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
15516 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
15517 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
15518 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15519 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15520 rather than TRULY_NOOP_TRUNCATION in comments.
15521 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
15522 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
15523 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
15524 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
15525 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
15526 * system.h (TRULY_NOOP_TRUNCATION): Poison.
15528 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
15531 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
15532 (*cmp_ior): Likewise.
15533 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
15534 (*ior_scc_scc_cmp): Likewise.
15535 (*and_scc_scc): Likewise.
15536 (*and_scc_scc_cmp): Likewise.
15538 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15539 Alan Hayard <alan.hayward@arm.com>
15540 David Sherwood <david.sherwood@arm.com>
15542 * target.def (can_change_mode_class): New hook.
15543 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
15544 (hard_regno_nregs): Likewise.
15545 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
15546 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
15547 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
15548 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
15549 (LOAD_EXTEND_OP): Update accordingly.
15550 * doc/tm.texi: Regenerate.
15551 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
15552 CANNOT_CHANGE_MODE_CLASS.
15553 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
15554 (REG_CAN_CHANGE_MODE_P): ...this new macro.
15555 * combine.c (simplify_set): Update accordingly.
15556 * emit-rtl.c (validate_subreg): Likewise.
15557 * recog.c (general_operand): Likewise.
15558 * regcprop.c (mode_change_ok): Likewise.
15559 * reload1.c (choose_reload_regs): Likewise.
15560 (inherit_piecemeal_p): Likewise.
15561 * rtlanal.c (simplify_subreg_regno): Likewise.
15562 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
15563 instead of CANNOT_CHANGE_MODE_CLASS.
15564 (reload_cse_simplify_operands): Likewise.
15565 * reload.c (push_reload): Use targetm.can_change_mode_class
15566 instead of CANNOT_CHANGE_MODE_CLASS.
15567 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
15568 REG_CANNOT_CHANGE_MODE_P.
15569 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15570 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
15571 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15572 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15573 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15574 (arm_can_change_mode_class): New function.
15575 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15576 than CANNOT_CHANGE_MODE_CLASS in comments.
15577 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15578 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
15579 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
15580 (ix86_can_change_mode_class): ...this new function, inverting the
15581 sense of the return value.
15582 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15583 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15584 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15585 (ia64_can_change_mode_class): New function.
15586 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15587 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
15588 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
15589 (m32c_can_change_mode_class): ...this new function, inverting the
15590 sense of the return value.
15591 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15592 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15593 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
15594 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
15595 (mips_can_change_mode_class): ...this new function, inverting the
15596 sense of the return value.
15597 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15598 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15599 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15600 (msp430_can_change_mode_class): New function.
15601 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15602 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
15603 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15604 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15605 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15606 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
15607 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15608 (pa_cannot_change_mode_class): Replace with...
15609 (pa_can_change_mode_class): ...this new function, inverting the
15610 sense of the return value.
15611 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15612 than CANNOT_CHANGE_MODE_CLASS in comments.
15613 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15614 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
15615 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15616 (pdp11_cannot_change_mode_class): Replace with...
15617 (pdp11_can_change_mode_class): ...this new function, inverting the
15618 sense of the return value.
15619 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15620 * config/powerpcspe/powerpcspe-protos.h
15621 (rs6000_cannot_change_mode_class_ptr): Delete.
15622 * config/powerpcspe/powerpcspe.c
15623 (rs6000_cannot_change_mode_class_ptr): Delete.
15624 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15625 (rs6000_option_override_internal): Assign to
15626 targetm.can_change_mode_class instead of
15627 rs6000_cannot_change_mode_class_ptr.
15628 (rs6000_cannot_change_mode_class): Replace with...
15629 (rs6000_can_change_mode_class): ...this new function, inverting the
15630 sense of the return value.
15631 (rs6000_debug_cannot_change_mode_class): Replace with...
15632 (rs6000_debug_can_change_mode_class): ...this new function.
15633 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15634 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
15635 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15636 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15637 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
15639 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
15640 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15641 (rs6000_option_override_internal): Assign to
15642 targetm.can_change_mode_class instead of
15643 rs6000_cannot_change_mode_class_ptr.
15644 (rs6000_cannot_change_mode_class): Replace with...
15645 (rs6000_can_change_mode_class): ...this new function, inverting the
15646 sense of the return value.
15647 (rs6000_debug_cannot_change_mode_class): Replace with...
15648 (rs6000_debug_can_change_mode_class): ...this new function.
15649 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15650 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
15651 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
15652 (s390_can_change_mode_class): ...this new function, inverting the
15653 sense of the return value.
15654 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15655 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15656 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
15657 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15658 (sh_cannot_change_mode_class): Replace with...
15659 (sh_can_change_mode_class): ...this new function, inverting the
15660 sense of the return value.
15661 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15662 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15663 (sparc_can_change_mode_class): New function.
15664 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15665 * config/spu/spu.c (spu_can_change_mode_class): New function.
15666 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15667 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15668 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15669 (visium_can_change_mode_class): New function.
15670 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
15672 2017-09-15 Richard Biener <rguenther@suse.de>
15674 PR tree-optimization/82217
15675 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
15676 but not undefined case.
15678 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15681 * postreload.c (reload_cse_simplify_operands): Skip
15682 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
15684 2017-09-15 Richard Biener <rguenther@suse.de>
15686 PR tree-optimization/68823
15687 * graphite-scop-detection.c (build_alias_set): If we have a
15688 possible dependence check whether we can handle them by just
15689 looking at the DRs DR_ACCESS_FNs.
15690 (build_scops): If build_alias_set fails, fail the SCOP.
15692 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15694 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
15695 to support float128 built-in functions that require the ISA 3.0
15697 (BU_FLOAT128_3_HW): Likewise.
15698 (SQRTF128): Add support for the IEEE 128-bit square root and fma
15699 built-in functions.
15700 (FMAF128): Likewise.
15702 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15703 support for built-in functions that need the ISA 3.0 IEEE 128-bit
15704 floating point instructions.
15705 (rs6000_invalid_builtin): Likewise.
15706 (rs6000_builtin_mask_names): Likewise.
15707 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
15708 (RS6000_BTM_FLOAT128_HW): Likewise.
15709 (RS6000_BTM_COMMON): Likewise.
15710 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
15712 * doc/extend.texi (RS/6000 built-in functions): Document the
15713 IEEE 128-bit floating point square root and fused multiply-add
15714 built-in functions.
15716 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
15718 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
15719 reg (r2) isn't in the set of registers defined in the prologue.
15721 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15722 Alan Hayward <alan.hayward@arm.com>
15723 David Sherwood <david.sherwood@arm.com>
15725 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
15726 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
15727 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
15728 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
15729 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
15731 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
15732 max_vectorization_factor.
15733 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
15735 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15736 Alan Hayward <alan.hayward@arm.com>
15737 David Sherwood <david.sherwood@arm.com>
15739 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
15740 (vect_worthwhile_without_simd_p): Declare.
15741 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
15742 (vectorizable_reduction): Use it.
15743 * tree-vect-stmts.c (vectorizable_shift): Likewise.
15744 (vectorizable_operation): Likewise.
15746 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15747 Alan Hayward <alan.hayward@arm.com>
15748 David Sherwood <david.sherwood@arm.com>
15750 * tree-vectorizer.h (vect_get_num_copies): New function.
15751 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
15752 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15753 (vectorizable_induction): Likewise.
15754 (vectorizable_live_operation): Likewise.
15755 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15756 (vectorizable_bswap): Likewise.
15757 (vectorizable_call): Likewise.
15758 (vectorizable_conversion): Likewise.
15759 (vectorizable_assignment): Likewise.
15760 (vectorizable_shift): Likewise.
15761 (vectorizable_operation): Likewise.
15762 (vectorizable_store): Likewise.
15763 (vectorizable_load): Likewise.
15764 (vectorizable_condition): Likewise.
15765 (vectorizable_comparison): Likewise.
15766 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
15768 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15769 Alan Hayward <alan.hayward@arm.com>
15770 David Sherwood <david.sherwood@arm.com>
15772 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
15773 of vect_init_vector.
15775 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15776 Alan Hayward <alan.hayward@arm.com>
15777 David Sherwood <david.sherwood@arm.com>
15779 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
15780 an inline wrapper that provides a location.
15781 (gimple_build_vector): Likewise.
15782 * gimple-fold.c (gimple_build_vector_from_val): New function.
15783 (gimple_build_vector): Likewise.
15784 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
15785 functions to build the initial value. Always return a gimple value.
15786 (get_initial_defs_for_reduction): Likewise. Only compute
15788 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
15789 vect_init_vector on the results from get_initial_def(s)_for_reduction.
15790 (vectorizable_induction): Use gimple_build_vector rather than
15793 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15794 Alan Hayward <alan.hayward@arm.com>
15795 David Sherwood <david.sherwood@arm.com>
15797 * target.h (vec_perm_indices): New typedef.
15798 (auto_vec_perm_indices): Likewise.
15799 * optabs-query.h: Include target.h
15800 (can_vec_perm_p): Take a vec_perm_indices *.
15801 * optabs-query.c (can_vec_perm_p): Likewise.
15802 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
15803 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15804 * tree-vect-generic.c (lower_vec_perm): Likewise.
15805 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15806 (vect_grouped_load_supported): Likewise.
15807 (vect_shift_permute_load_chain): Likewise.
15808 (vect_permute_store_chain): Use auto_vec_perm_indices.
15809 (vect_permute_load_chain): Likewise.
15810 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15811 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15812 Update uses of can_vec_perm_p.
15813 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15814 mode with a number of elements. Take a vec_perm_indices *.
15815 (vect_create_epilog_for_reduction): Update accordingly.
15816 Use auto_vec_perm_indices.
15817 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15818 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15819 (vect_transform_slp_perm_load): Likewise.
15820 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15821 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15822 (vect_gen_perm_mask_checked): Likewise.
15823 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15824 (vect_gen_perm_mask_checked): Likewise.
15825 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15826 (vectorizable_store): Likewise.
15827 (vectorizable_load): Likewise.
15828 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15829 (vectorizable_bswap): Likewise.
15831 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15832 Alan Hayward <alan.hayward@arm.com>
15833 David Sherwood <david.sherwood@arm.com>
15835 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15836 * tree.c (build_vector): Likewise.
15837 (build_vector_from_ctor): Update accordingly.
15838 (build_vector_from_val): Likewise.
15839 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15840 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15841 * tree-vect-generic.c (add_rshift): Likewise.
15842 (expand_vector_divmod): Likewise.
15843 (optimize_vector_constructor): Likewise.
15844 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15845 (vect_transform_slp_perm_load): Likewise.
15846 (vect_schedule_slp_instance): Likewise.
15847 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15848 (vectorizable_call): Likewise.
15849 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15850 * expmed.c (make_tree): Likewise.
15851 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15852 a vector passed to build_vector.
15853 (fold_convert_const): Likewise.
15854 (exact_inverse): Likewise.
15855 (fold_ternary_loc): Likewise.
15856 (fold_relational_const): Likewise.
15857 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15858 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15859 (const_unop): Likewise. Store the reduction accumulator in a
15860 variable rather than an array.
15861 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15862 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15863 the new vector, rather than constructing it after the input arrays.
15864 (native_interpret_vector): Use auto_vec<tree> when building
15865 a vector passed to build_vector. Add elements in order.
15866 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15867 auto_vec<tree> when building a vector passed to build_vector.
15868 (vect_create_epilog_for_reduction): Likewise.
15869 (vectorizable_induction): Likewise.
15870 (get_initial_def_for_reduction): Likewise. Fix indentation of
15872 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15874 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15875 passed to build_vector.
15877 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15878 Alan Hayward <alan.hayward@arm.com>
15879 David Sherwood <david.sherwood@arm.com>
15881 * tree-core.h (tree_base::u): Add an "nelts" field.
15882 (tree_vector): Use VECTOR_CST_NELTS as the length.
15883 * tree.c (tree_size): Likewise.
15884 (make_vector): Initialize VECTOR_CST_NELTS.
15885 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15886 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15887 TYPE_VECTOR_SUBPARTS.
15888 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15889 as the number of units, setting it from VECTOR_CST_NELTS.
15890 (const_vector_from_tree): Likewise.
15891 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15892 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15893 (fold_negate_expr_1): Likewise.
15894 (fold_convert_const): Likewise.
15895 (const_binop): Likewise. Differentiate the number of output and
15897 (const_unop): Likewise.
15898 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15899 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15900 in cases that did the opposite.
15902 2017-09-14 Richard Biener <rguenther@suse.de>
15904 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15907 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15909 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15911 2017-09-14 Jakub Jelinek <jakub@redhat.com>
15914 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
15915 if and where to split a bb, except for splitting before debug insn
15916 sequences followed by non-label real insn. Delete debug insns
15917 in between basic blocks.
15919 * combine.c (make_compound_operation_int): Formatting fixes.
15921 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
15922 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15923 * config/netbsd.h (LINK_EH_SPEC): Likewise.
15924 * config/sol2.h (LINK_EH_SPEC): Likewise.
15925 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15926 * config/s390/linux.h (LINK_SPEC): Likewise.
15927 * config/freebsd.h (LINK_EH_SPEC): Likewise.
15928 * config/openbsd.h (LINK_EH_SPEC): Likewise.
15929 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15930 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
15931 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
15932 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15933 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
15934 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
15936 2017-09-13 Jakub Jelinek <jakub@redhat.com>
15938 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
15940 (ENDFILE_LINUX_SPEC): Likewise.
15941 (LINK_EH_SPEC): Likewise.
15942 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
15943 (LINK_OS_LINUX_SPEC32): Likewise.
15944 (LINK_OS_LINUX_SPEC64): Likewise.
15945 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
15946 (LINK_OS_LINUX_SPEC): Likewise.
15948 2017-09-13 Martin Liska <mliska@suse.cz>
15950 PR middle-end/82154
15951 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
15952 CASE_HIGH is NULL_TREE.
15954 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15955 Alan Hayward <alan.hayward@arm.com>
15956 David Sherwood <david.sherwood@arm.com>
15958 * target.def (secondary_memory_needed): New hook.
15959 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
15960 instead of SECONDARY_MEMORY_NEEDED.
15961 (secondary_memory_needed_mode): Likewise.
15962 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
15963 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
15964 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
15965 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
15966 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
15967 * doc/tm.texi: Regenerate.
15968 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
15969 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
15970 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15971 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
15972 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
15973 * config/i386/i386.c (inline_secondary_memory_needed): Put the
15974 mode argument first and change the reg_class arguments to reg_class_t.
15975 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
15976 Make static. Update the call to inline_secondary_memory_needed.
15977 (ix86_register_move_cost): Update the call to
15978 inline_secondary_memory_needed.
15979 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15980 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
15982 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
15983 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15985 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
15986 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
15987 * config/mips/mips.c (mips_secondary_memory_needed): Make static
15988 and match hook interface. Add comment from mips.h.
15989 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15990 * config/mmix/mmix.md (truncdfsf2): Refer to
15991 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15993 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
15994 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
15995 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15996 (pa_secondary_memory_needed): New function.
15997 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
15998 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
15999 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16000 (pdp11_secondary_memory_needed): Make static and match hook interface.
16001 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
16002 * config/powerpcspe/powerpcspe-protos.h
16003 (rs6000_secondary_memory_needed_ptr): Delete.
16004 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
16006 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16007 (rs6000_option_override_internal): Assign to
16008 targetm.secondary_memory_needed rather than
16009 rs6000_secondary_memory_needed_ptr.
16010 (rs6000_secondary_memory_needed): Match hook interface.
16011 (rs6000_debug_secondary_memory_needed): Likewise.
16012 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
16013 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
16014 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
16015 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16016 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
16017 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
16019 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
16020 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16021 (rs6000_option_override_internal): Assign to
16022 targetm.secondary_memory_needed rather than
16023 rs6000_secondary_memory_needed_ptr.
16024 (rs6000_secondary_memory_needed): Match hook interface.
16025 (rs6000_debug_secondary_memory_needed): Likewise.
16026 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
16027 * config/s390/s390.c (s390_secondary_memory_needed): New function.
16028 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16029 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
16030 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16031 (sparc_secondary_memory_needed): New function.
16032 * lra-constraints.c (check_and_process_move): Refer to
16033 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
16035 (curr_insn_transform): Likewise.
16036 (process_alt_operands): Use targetm.secondary_memory_needed
16037 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16038 (check_secondary_memory_needed_p): Likewise.
16039 (choose_split_class): Likewise.
16040 * reload.c: Unconditionally include code that was previously
16041 conditional on SECONDARY_MEMORY_NEEDED.
16042 (push_secondary_reload): Use targetm.secondary_memory_needed
16043 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16044 (push_reload): Likewise.
16045 * reload1.c: Unconditionally include code that was previously
16046 conditional on SECONDARY_MEMORY_NEEDED.
16047 (choose_reload_regs): Use targetm.secondary_memory_needed
16048 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16049 (gen_reload): Likewise.
16050 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
16052 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
16053 Alan Hayward <alan.hayward@arm.com>
16054 David Sherwood <david.sherwood@arm.com>
16056 * target.def (secondary_memory_needed_mode): New hook:
16057 * targhooks.c (default_secondary_memory_needed_mode): Declare.
16058 * targhooks.h (default_secondary_memory_needed_mode): New function.
16059 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
16060 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
16061 * doc/tm.texi: Regenerate.
16062 * lra-constraints.c (check_and_process_move): Use
16063 targetm.secondary_memory_needed_mode instead of
16064 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
16065 (curr_insn_transform): Likewise.
16066 * reload.c (get_secondary_mem): Likewise.
16067 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16068 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
16070 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16071 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16072 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
16073 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16074 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
16076 * config/powerpcspe/powerpcspe-protos.h
16077 (rs6000_secondary_memory_needed_mode): Delete.
16078 * config/powerpcspe/powerpcspe.c
16079 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16080 (rs6000_secondary_memory_needed_mode): Make static.
16081 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16082 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
16084 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16086 (rs6000_secondary_memory_needed_mode): Make static.
16087 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16088 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
16089 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16090 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16091 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16093 (sparc_secondary_memory_needed_mode): New function.
16094 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
16096 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
16098 * config/aarch64/constraints.md (Umq): New constraint.
16099 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
16101 (mov<mode>): Update condition.
16103 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16105 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
16106 when bitposition is the same.
16108 2017-09-13 Richard Biener <rguenther@suse.de>
16110 * dwarf2out.c (output_die_symbol): Remove.
16111 (output_die): Do not output a DIEs symbol.
16113 2017-09-13 Richard Biener <rguenther@suse.de>
16115 PR middle-end/82128
16116 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
16117 default-def to avoid breaking iterator update with the weird
16118 interaction with cgraph_update_edges_for_call_stmt_node.
16120 2017-09-13 Richard Biener <rguenther@suse.de>
16122 * tree-cfg.c (verify_gimple_assign_binary): Add verification
16123 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
16124 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
16125 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
16127 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
16129 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
16130 Disable pc relative literal load irrespective of
16131 TARGET_FIX_ERR_A53_84341 for default.
16133 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
16135 * config/sparc/sparc.c (output_return): Output the source location of
16136 the insn in the delay slot, if any.
16137 (output_sibcall): Likewise.
16139 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
16142 * common.opt (-static-pie): New alias.
16143 (shared): Negate static-pie.
16144 (-no-pie): Update help text.
16146 (static-pie): New option.
16147 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
16148 -static-pie support.
16149 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
16150 (LINK_EH_SPEC): Likewise.
16151 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16152 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16153 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16154 * gcc.c (LINK_COMMAND_SPEC): Likewise.
16155 (init_gcc_specs): Likewise.
16156 (init_spec): Likewise.
16157 (display_help): Update help message for -pie.
16158 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
16161 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
16163 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
16164 (movdi_aarch64): Likewise.
16165 (movti_aarch64): Likewise.
16167 2017-09-12 Simon Wright <simon@pushface.org>
16170 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
16171 calculation of the minor version, always output as 0.
16173 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16176 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
16177 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
16178 on it early, rather than manual conversion late. For
16179 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
16180 instead of performing manual conversion.
16182 2017-09-12 Carl Love <cel@us.ibm.com>
16184 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
16185 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
16187 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16188 VMULOSW): Add definitions.
16189 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16190 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16191 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16192 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16193 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16195 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16197 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
16199 (movti_aarch64): Likewise.
16200 (movdf_aarch64): Likewise.
16201 (movtf_aarch64): Likewise.
16202 (load_pairdi): Likewise.
16203 (store_pairdi): Likewise.
16204 (load_pairdf): Likewise.
16205 (store_pairdf): Likewise.
16206 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
16207 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
16208 (ldr_got_small_<mode>): Likewise.
16209 (ldr_got_small_28k_<mode>): Likewise.
16210 (ldr_got_tiny): Likewise.
16211 * config/aarch64/iterators.md (ldst_sz): New.
16212 (ldpstp_sz): Likewise.
16213 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
16215 (thunderx_load): Split load_8 to load_16.
16216 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
16218 (thunderx2t99_storepair_basic): Split store_8 to store_16.
16219 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
16220 (xgene1_store_pair): Split store_8 to store_16.
16221 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
16222 (falkor_st_0_st_sd): Split store_8 to store_16.
16224 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16226 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
16227 and store1/2/3/4 to store_4/8/12/16.
16228 * config/aarch64/aarch64.md: Update for rename.
16229 * config/arm/arm.md: Likewise.: Likewise.
16230 * config/arm/arm.c: Likewise.
16231 * config/arm/thumb1.md: Likewise.
16232 * config/arm/thumb2.md: Likewise.
16233 * config/arm/vfp.md: Likewise.
16234 * config/arm/arm-generic.md: Likewise.
16235 * config/arm/arm1020e.md: Likewise.
16236 * config/arm/arm1026ejs.md: Likewise.
16237 * config/arm/arm1136jfs.md: Likewise.
16238 * config/arm/arm926ejs.md: Likewise.
16239 * config/arm/cortex-a15.md: Likewise.
16240 * config/arm/cortex-a17.md: Likewise.
16241 * config/arm/cortex-a5.md: Likewise.
16242 * config/arm/cortex-a53.md: Likewise.
16243 * config/arm/cortex-a57.md: Likewise.
16244 * config/arm/cortex-a7.md: Likewise.
16245 * config/arm/cortex-a8.md: Likewise.
16246 * config/arm/cortex-a9.md: Likewise.
16247 * config/arm/cortex-m4.md: Likewise.
16248 * config/arm/cortex-m7.md: Likewise.
16249 * config/arm/cortex-r4.md: Likewise.
16250 * config/arm/exynos-m1.md: Likewise.
16251 * config/arm/fa526.md: Likewise.
16252 * config/arm/fa606te.md: Likewise.
16253 * config/arm/fa626te.md: Likewise.
16254 * config/arm/fa726te.md: Likewise.
16255 * config/arm/fmp626.md: Likewise.
16256 * config/arm/iwmmxt.md: Likewise.
16257 * config/arm/ldmstm.md: Likewise.
16258 * config/arm/marvell-pj4.md: Likewise.
16259 * config/arm/xgene1.md: Likewise.
16260 * config/aarch64/thunderx.md: Likewise.
16261 * config/aarch64/thunderx2t99.md: Likewise.
16262 * config/aarch64/falkor.md: Likewise.
16264 2017-09-12 Martin Liska <mliska@suse.cz>
16266 * attribs.c (private_lookup_attribute): New function.
16267 * attribs.h (private_lookup_attribute): Declared here.
16268 (lookup_attribute): Called from this place.
16270 2017-09-12 Richard Biener <rguenther@suse.de>
16272 PR tree-optimization/82157
16273 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
16274 stmts with side-effects.
16276 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16277 Alan Hayward <alan.hayward@arm.com>
16278 David Sherwood <david.sherwood@arm.com>
16280 * target.def (hard_regno_nregs): New hook.
16281 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
16282 * targhooks.h (default_hard_regno_nregs): Declare.
16283 * targhooks.c (default_hard_regno_nregs): New function.
16284 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
16285 (TARGET_HARD_REGNO_NREGS): ...this hook.
16286 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
16287 (CLASS_MAX_NREGS): Likewise.
16288 * doc/tm.texi: Regenerate.
16289 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
16290 instead of HARD_REGNO_NREGS.
16291 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
16292 HARD_REGNO_NREGS in the comment.
16293 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
16294 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
16295 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
16296 Return an unsigned int.
16297 (TARGET_HARD_REGNO_NREGS): Redefine.
16298 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
16299 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
16300 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16301 (arc_hard_regno_nregs): New function.
16302 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
16303 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
16304 (arm_hard_regno_nregs): New function.
16305 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
16306 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
16307 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
16308 (TARGET_HARD_REGNO_NREGS): Redefine.
16309 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
16310 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
16311 (HARD_REGNO_NREGS): Delete.
16312 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
16313 (cr16_hard_regno_nregs): New function.
16314 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
16315 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
16316 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
16317 (cris_hard_regno_nregs): New function.
16318 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
16319 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
16320 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16321 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
16322 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
16323 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
16324 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
16325 (frv_hard_regno_nregs): Make static. Take and return an
16327 (frv_class_max_nregs): Remove outdated copy of documentation.
16328 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
16329 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
16330 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
16331 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
16332 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
16333 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
16334 (TARGET_HARD_REGNO_NREGS): Redefine.
16335 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
16336 (CLASS_MAX_NREGS): Update comment.
16337 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
16338 (ia64_hard_regno_nregs): New function.
16339 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
16340 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
16341 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
16342 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
16343 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
16345 (m32c_hard_regno_nregs): Likewise. Make static.
16346 (TARGET_HARD_REGNO_NREGS): Redefine.
16347 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
16348 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
16349 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
16350 (m68k_hard_regno_nregs): New function.
16351 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
16352 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
16353 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
16354 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
16355 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
16356 Take and return an unsigned int.
16357 (TARGET_HARD_REGNO_NREGS): Redefine.
16358 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
16359 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16360 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
16361 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
16362 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
16363 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
16364 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
16365 (msp430_hard_regno_nregs): Make static. Take and return an
16367 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
16368 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
16369 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
16370 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
16371 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
16372 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
16373 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
16374 (TARGET_HARD_REGNO_NREGS): Redefine.
16375 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
16376 (PA_HARD_REGNO_NREGS): ...this.
16377 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
16378 (PA_HARD_REGNO_NREGS): ...this.
16379 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16380 (pa_hard_regno_nregs): New function.
16381 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
16382 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
16383 (pdp11_hard_regno_nregs): New function.
16384 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
16385 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
16386 (rs6000_hard_regno_nregs_hook): New function.
16387 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
16388 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
16389 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
16390 Take and return an unsigned int. Move earlier in file.
16391 (TARGET_HARD_REGNO_NREGS): Redefine.
16392 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
16393 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
16394 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
16395 (rl78_hard_regno_nregs): Make static. Take and return an
16397 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
16398 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
16399 (rs6000_hard_regno_nregs_hook): New function.
16400 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
16401 * config/rx/rx.c (rx_hard_regno_nregs): New function.
16402 (TARGET_HARD_REGNO_NREGS): Redefine.
16403 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
16404 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
16405 instead of HARD_REGNO_NREGS.
16406 (s390_hard_regno_nregs): New function.
16407 (s390_hard_regno_mode_ok): Add comment from s390.h.
16408 (TARGET_HARD_REGNO_NREGS): Redefine.
16409 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
16410 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
16411 (sh_hard_regno_nregs): New function.
16412 (sh_pass_in_reg_p): Use it.
16413 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
16414 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16415 (sparc_hard_regno_nregs): New function.
16416 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
16417 * config/spu/spu.c (spu_hard_regno_nregs): New function.
16418 (spu_function_arg_advance): Use it, supplying a valid register number.
16419 (TARGET_HARD_REGNO_NREGS): Redefine.
16420 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
16421 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
16422 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
16423 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
16424 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
16425 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
16426 (CLASS_MAX_NREGS): Remove copy of old documentation.
16427 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
16428 (visium_hard_regno_nregs): New function.
16429 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
16430 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
16431 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16432 xtensa_hard_regno_nregs): New function.
16433 * system.h (HARD_REGNO_NREGS): Poison.
16435 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16437 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
16438 hard_regno_nregs instead of HARD_REGNO_NREGS.
16439 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
16440 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
16441 (c6x_expand_epilogue): Likewise.
16442 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
16443 (frv_read_iacc_argument): Likewise.
16444 * config/sh/sh.c: Include regs.h.
16445 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16446 (regs_used): Likewise.
16447 (output_stack_adjust): Likewise.
16448 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
16449 * expmed.c: Include regs.h.
16450 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16451 * ree.c: Include regs.h.
16452 (combine_reaching_defs): Use hard_regno_nregs instead of
16454 (add_removable_extension): Likewise.
16456 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16458 * regs.h (hard_regno_nregs): Turn into a function.
16459 (end_hard_regno): Update accordingly.
16460 * caller-save.c (setup_save_areas): Likewise.
16461 (save_call_clobbered_regs): Likewise.
16462 (replace_reg_with_saved_mem): Likewise.
16463 (insert_restore): Likewise.
16464 (insert_save): Likewise.
16465 * combine.c (can_change_dest_mode): Likewise.
16466 (move_deaths): Likewise.
16467 (distribute_notes): Likewise.
16468 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
16469 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
16470 (rs6000_split_multireg_move): Likewise.
16471 (rs6000_register_move_cost): Likewise.
16472 (rs6000_memory_move_cost): Likewise.
16473 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
16474 (rs6000_split_multireg_move): Likewise.
16475 (rs6000_register_move_cost): Likewise.
16476 (rs6000_memory_move_cost): Likewise.
16477 * cselib.c (cselib_reset_table): Likewise.
16478 (cselib_lookup_1): Likewise.
16479 * emit-rtl.c (set_mode_and_regno): Likewise.
16480 * function.c (aggregate_value_p): Likewise.
16481 * ira-color.c (setup_profitable_hard_regs): Likewise.
16482 (check_hard_reg_p): Likewise.
16483 (calculate_saved_nregs): Likewise.
16484 (assign_hard_reg): Likewise.
16485 (improve_allocation): Likewise.
16486 (calculate_spill_cost): Likewise.
16487 * ira-emit.c (modify_move_list): Likewise.
16488 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
16489 (ira_hard_reg_in_set_p): Likewise.
16490 * ira.c (setup_reg_mode_hard_regset): Likewise.
16491 (clarify_prohibited_class_mode_regs): Likewise.
16492 (check_allocation): Likewise.
16493 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16494 (lra_setup_reg_renumber): Likewise.
16495 (setup_try_hard_regno_pseudos): Likewise.
16496 (spill_for): Likewise.
16497 (assign_hard_regno): Likewise.
16498 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
16499 * lra-constraints.c (in_class_p): Likewise.
16500 (lra_constraint_offset): Likewise.
16501 (simplify_operand_subreg): Likewise.
16502 (lra_constraints): Likewise.
16503 (split_reg): Likewise.
16504 (split_if_necessary): Likewise.
16505 (invariant_p): Likewise.
16506 (inherit_in_ebb): Likewise.
16507 * lra-lives.c (process_bb_lives): Likewise.
16508 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
16509 (get_hard_regs): Likewise.
16510 (do_remat): Likewise.
16511 * lra-spills.c (assign_spill_hard_regs): Likewise.
16512 * mode-switching.c (create_pre_exit): Likewise.
16513 * postreload.c (reload_combine_recognize_pattern): Likewise.
16514 * recog.c (peep2_find_free_register): Likewise.
16515 * regcprop.c (kill_value_regno): Likewise.
16516 (set_value_regno): Likewise.
16517 (copy_value): Likewise.
16518 (maybe_mode_change): Likewise.
16519 (find_oldest_value_reg): Likewise.
16520 (copyprop_hardreg_forward_1): Likewise.
16521 * regrename.c (check_new_reg_p): Likewise.
16522 (regrename_do_replace): Likewise.
16523 * reload.c (push_reload): Likewise.
16524 (combine_reloads): Likewise.
16525 (find_dummy_reload): Likewise.
16526 (operands_match_p): Likewise.
16527 (find_reloads): Likewise.
16528 (find_equiv_reg): Likewise.
16529 (reload_adjust_reg_for_mode): Likewise.
16530 * reload1.c (count_pseudo): Likewise.
16531 (count_spilled_pseudo): Likewise.
16532 (find_reg): Likewise.
16533 (clear_reload_reg_in_use): Likewise.
16534 (free_for_value_p): Likewise.
16535 (allocate_reload_reg): Likewise.
16536 (choose_reload_regs): Likewise.
16537 (reload_adjust_reg_for_temp): Likewise.
16538 (emit_reload_insns): Likewise.
16539 (delete_output_reload): Likewise.
16540 * rtlanal.c (subreg_get_info): Likewise.
16541 * sched-deps.c (sched_analyze_reg): Likewise.
16542 * sel-sched.c (init_regs_for_mode): Likewise.
16543 (mark_unavailable_hard_regs): Likewise.
16544 (choose_best_reg_1): Likewise.
16545 (verify_target_availability): Likewise.
16546 * valtrack.c (dead_debug_insert_temp): Likewise.
16547 * var-tracking.c (track_loc_p): Likewise.
16548 (emit_note_insn_var_location): Likewise.
16549 * varasm.c (make_decl_rtl): Likewise.
16550 * reginfo.c (choose_hard_reg_mode): Likewise.
16551 (init_reg_modes_target): Refer directly to
16552 this_target_regs->x_hard_regno_nregs.
16554 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16556 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
16557 instead of hard_regno_nregs.
16559 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16561 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
16562 end_hard_regno instead of hard_regno_nregs.
16563 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
16564 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
16565 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
16566 * ira-color.c (improve_allocation): Likewise.
16567 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16568 * lra-lives.c (mark_regno_live): Likewise.
16569 (mark_regno_dead): Likewise.
16570 * lra-remat.c (operand_to_remat): Likewise.
16571 * lra.c (collect_non_operand_hard_regs): Likewise.
16572 * postreload.c (reload_combine_note_store): Likewise.
16573 (move2add_valid_value_p): Likewise.
16574 * reload.c (regno_clobbered_p): Likewise.
16576 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16578 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
16580 * config/v850/v850.c (v850_reorg): Likewise.
16581 * reload.c (refers_to_regno_for_reload_p): Likewise.
16582 (find_equiv_reg): Likewise.
16583 * reload1.c (reload_reg_reaches_end_p): Likewise.
16585 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16587 * caller-save.c (add_used_regs): Use REG_NREGS instead of
16589 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
16590 * config/arm/arm.c (output_move_neon): Likewise.
16591 (arm_attr_length_move_neon): Likewise.
16592 (neon_split_vcombine): Likewise.
16593 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
16594 (c6x_mark_reg_written): Likewise.
16595 (c6x_dwarf_register_span): Likewise.
16596 * config/i386/i386.c (ix86_save_reg): Likewise.
16597 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
16598 (rws_access_reg): Likewise.
16599 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
16600 * mode-switching.c (create_pre_exit): Likewise.
16601 * ree.c (combine_reaching_defs): Likewise.
16602 (add_removable_extension): Likewise.
16603 * regcprop.c (find_oldest_value_reg): Likewise.
16604 (copyprop_hardreg_forward_1): Likewise.
16605 * reload.c (reload_inner_reg_of_subreg): Likewise.
16606 (push_reload): Likewise.
16607 (combine_reloads): Likewise.
16608 (find_dummy_reload): Likewise.
16609 (reload_adjust_reg_for_mode): Likewise.
16610 * reload1.c (find_reload_regs): Likewise.
16611 (forget_old_reloads_1): Likewise.
16612 (reload_reg_free_for_value_p): Likewise.
16613 (reload_adjust_reg_for_temp): Likewise.
16614 (emit_reload_insns): Likewise.
16615 (delete_output_reload): Likewise.
16616 * sel-sched.c (choose_best_reg_1): Likewise.
16617 (choose_best_pseudo_reg): Likewise.
16619 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16620 Alan Hayward <alan.hayward@arm.com>
16621 David Sherwood <david.sherwood@arm.com>
16623 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
16624 * target.def (slow_unaligned_access): New hook.
16625 * targhooks.h (default_slow_unaligned_access): Declare.
16626 * targhooks.c (default_slow_unaligned_access): New function.
16627 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
16628 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
16629 * doc/tm.texi: Regenerate.
16630 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
16631 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
16632 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
16634 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
16635 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
16637 (rs6000_slow_unaligned_access): New function.
16638 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16639 (expand_block_compare): Likewise.
16640 (expand_strn_compare): Likewise.
16641 (rs6000_rtx_costs): Likewise.
16642 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
16643 (riscv_slow_unaligned_access): Likewise.
16644 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
16645 (riscv_slow_unaligned_access_p): ...this and make static.
16646 (riscv_option_override): Update accordingly.
16647 (riscv_slow_unaligned_access): New function.
16648 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16649 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
16650 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16651 (rs6000_slow_unaligned_access): New function.
16652 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16653 (rs6000_rtx_costs): Likewise.
16654 * config/rs6000/rs6000-string.c (expand_block_compare)
16655 (expand_strn_compare): Use targetm.slow_unaligned_access instead
16656 of SLOW_UNALIGNED_ACCESS.
16657 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
16658 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
16659 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
16660 of SLOW_UNALIGNED_ACCESS.
16661 * expmed.c (simple_mem_bitfield_p): Likewise.
16662 * expr.c (alignment_for_piecewise_move): Likewise.
16663 (emit_group_load_1): Likewise.
16664 (emit_group_store): Likewise.
16665 (copy_blkmode_from_reg): Likewise.
16666 (emit_push_insn): Likewise.
16667 (expand_assignment): Likewise.
16668 (store_field): Likewise.
16669 (expand_expr_real_1): Likewise.
16670 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
16671 * lra-constraints.c (simplify_operand_subreg): Likewise.
16672 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16673 * gimple-ssa-store-merging.c: Likewise in block comment at start
16675 * tree-ssa-strlen.c: Include target.h.
16676 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
16677 of SLOW_UNALIGNED_ACCESS.
16678 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
16680 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16682 PR rtl-optimization/82185
16683 * expmed.c (emit_store_flag_int): Only test tem if it has been
16686 2017-09-12 Richard Biener <rguenther@suse.de>
16688 PR middle-end/82149
16689 * match.pd ((FTYPE) N CMP CST): Fix typo.
16691 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
16693 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
16695 (mips_near_type_p): Add 'short_call' attribute as a synonym
16697 * doc/extend.texi (short_call): Document new function attribute.
16699 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16702 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
16703 assertion check that in the condition.
16704 (get_atomic_generic_size): Likewise. Before testing if parameter
16705 has pointer type, if it has array type, call for C++
16706 default_conversion to perform array-to-pointer conversion.
16708 2017-09-12 Richard Biener <rguenther@suse.de>
16710 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
16711 for operations we cannot scalarize.
16713 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
16715 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
16716 vectors heap vectors. Clean up comments.
16717 Make visited_bbs a reference.
16718 (profitable_jump_thread_path): Make GC
16719 vectors heap vectors. Clean up comments.
16721 (convert_and_register_jump_thread_path): Make GC vectors heap
16723 (check_subpath_and_update_thread_path): Same. Clean up comments.
16724 Make visited_bbs a reference.
16725 (handle_phi): Abstract common code to to
16726 register_jump_thread_path_if_profitable.
16727 Rename VAR_BB to DEF_BB.
16729 Make GC vectors heap vectors.
16730 Make visited_bbs a reference.
16731 (handle_assignment): Same.
16732 (register_jump_thread_path_if_profitable): New.
16733 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
16735 Make GC vectors heap vectors. Clean up comments.
16736 Make visited_bbs a reference.
16737 (find_jump_threads_backwards): Make visited_bbs live in the stack.
16738 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
16741 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
16744 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
16745 words of E_DImode object are reachable by xtensa_uimm8x4 access.
16747 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
16749 Revert r251800 and r251799.
16751 2017-09-11 Martin Jambor <mjambor@suse.cz>
16754 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
16755 arguments in advance.
16756 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
16757 use it to find predecessor edges.
16758 (naive_outof_ssa): Collect vector of predecessors.
16760 2017-09-08 Jason Merrill <jason@redhat.com>
16762 PR c++/70029 - ICE with ref-qualifier and -flto
16763 * langhooks.h (struct lang_hooks_for_types): Add
16764 copy_lang_qualifiers.
16765 * attribs.c (build_type_attribute_qual_variant): Use it.
16766 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
16768 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
16769 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
16771 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
16774 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
16775 (*mulsi3_sp64): New instruction.
16776 (mulsi3): New expander.
16778 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
16780 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
16782 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16784 * sancov.c: Include memmodel.h.
16786 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
16789 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
16792 2017-09-07 Carl Love <cel@us.ibm.com>
16794 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
16795 the sldi instruction.
16797 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
16799 * sancov.c: Include tm_p.h.
16801 2017-09-07 Jakub Jelinek <jakub@redhat.com>
16804 * output.h (switch_to_other_text_partition): New declaration.
16805 * varasm.c (switch_to_other_text_partition): New function.
16806 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16807 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16808 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16809 to the other text partition before emitting LCL label and switch back
16810 after emitting the word after it.
16812 2017-09-07 Richard Biener <rguenther@suse.de>
16814 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16815 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16816 critical edges here, after loop init.
16817 (pass_data_pre): Remove PROP_no_crit_edges flags.
16818 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16819 for valueization of call args to avoid leaking VN_TOP.
16820 (visit_use): Assert we do not visit default defs.
16821 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16822 Use error_mark_node to more easily detect leaking VN_TOP.
16823 All default-defs are varying, not VN_TOP. Mark them visited.
16824 (run_scc_vn): Make code match comment.
16826 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16828 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16829 OPTION_MASK_FLOAT128_KEYWORD.
16830 (POWERPC_MASKS): Likewise.
16831 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16832 support for the -mfloat128-type option, and make -mfloat128
16833 default on PowerPC Linux systems. Define or undefine
16834 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16835 Define __float128 to be __ieee128 if IEEE 128-bit support is
16836 enabled, or undefine it.
16837 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16838 Delete defining __FLOAT128_TYPE__.
16839 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16840 -mfloat128-type option and make -mfloat128 default on PowerPC
16842 (TARGET_FLOAT128_TYPE): Likewise.
16843 (-mfloat128-type): Likewise.
16844 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16845 Delete the -mfloat128-type option and make -mfloat128 default on
16846 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16847 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16848 128-bit floating point is enabled. Change tests from using
16849 -mfloat128-type to -mfloat128.
16850 (rs6000_mangle_type): Use the correct mangling for the __float128
16851 type even if normal long double is restricted to 64-bits.
16852 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16854 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16855 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16856 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16858 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16859 documentation for -mfloat128.
16861 2017-09-06 Olivier Hainque <hainque@adacore.com>
16863 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16865 2017-09-06 Wish Wu <wishwu007@gmail.com>
16866 Jakub Jelinek <jakub@redhat.com>
16868 * asan.c (initialize_sanitizer_builtins): Add
16869 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16870 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16871 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16872 BT_FN_VOID_UINT64_PTR variables.
16873 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16874 (BT_FN_VOID_UINT16_UINT16): Likewise.
16875 (BT_FN_VOID_UINT32_UINT32): Likewise.
16876 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16877 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16878 (BT_FN_VOID_UINT64_PTR): Likewise.
16879 * common.opt (flag_sanitize_coverage): New variable.
16880 (fsanitize-coverage=trace-pc): Remove.
16881 (fsanitize-coverage=): Add.
16882 * flag-types.h (enum sanitize_coverage_code): New enum.
16883 * fold-const.c (fold_range_test): Disable non-short-circuit
16884 optimization if flag_sanitize_coverage.
16885 (fold_truth_andor): Likewise.
16886 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16887 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16888 (coverage_sanitizer_opts): New array.
16889 (get_closest_sanitizer_option): Add OPTS argument, handle also
16890 OPT_fsanitize_coverage_.
16891 (parse_sanitizer_options): Adjusted to also handle
16892 OPT_fsanitize_coverage_.
16893 (common_handle_option): Add OPT_fsanitize_coverage_.
16894 * sancov.c (instrument_comparison, instrument_switch): New function.
16895 (sancov_pass): Add trace-cmp support.
16896 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16897 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16898 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16899 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16900 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16901 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16902 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16903 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16904 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16905 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16907 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16909 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16910 error. Only quit immediately if parsing is complete.
16911 (BEGIN): Initialize fatal_err and parse_done.
16912 (begin fpu, end fpu): Check number of arguments.
16913 (begin arch, end arch): Likewise.
16914 (begin cpu, end cpu): Likewise.
16915 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
16916 (optalias): Likewise.
16918 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16920 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
16921 * config/arm/arm-isa.h: Delete. Move definitions to ...
16922 * arm-cpus.in: ... here. Use new feature and fgroup values.
16923 * config/arm/arm.c (arm_option_override): Use lower case for feature
16925 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
16926 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
16927 * config/arm/parsecpu.awk (END): Add new command 'isa'.
16929 (print_isa_bits_for): New function.
16930 (gen_isa): New function.
16931 (gen_comm_data): Use print_isa_bits_for.
16932 (define feature): New keyword.
16933 (define fgroup): New keyword.
16934 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
16935 (arm-isa.h): Add rule to generate file.
16936 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
16937 case for feature bit names.
16939 2017-09-06 Richard Biener <rguenther@suse.de>
16941 * tree-ssa-pre.c (NECESSARY): Remove.
16942 (create_expression_by_pieces): Do not touch pass-local flags.
16943 (insert_into_preds_of_block): Likewise.
16944 (do_pre_regular_insertion): Likewise.
16945 (eliminate_insert): Likewise.
16946 (eliminate_dom_walker::before_dom_children): Likewise.
16947 (fini_eliminate): Do not look at inserted_exprs.
16948 (mark_operand_necessary): Remove.
16949 (remove_dead_inserted_code): Replace with simple work-list
16950 algorithm based on inserted_exprs and SSA uses.
16951 (pass_pre::execute): Re-order fini_eliminate and
16952 remove_dead_inserted_code.
16954 2017-09-06 Olivier Hainque <hainque@adacore.com>
16956 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
16957 for VxWorks 7. Adjust surrounding comments.
16959 2017-09-06 Richard Biener <rguenther@suse.de>
16961 * gimple-ssa-strength-reduction.c
16962 (find_candidates_dom_walker::before_dom_children): Also allow
16965 2017-09-06 Richard Biener <rguenther@suse.de>
16967 PR tree-optimization/82108
16968 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
16969 for gap in the non-permutation SLP case.
16971 2017-09-06 Martin Jambor <mjambor@suse.cz>
16973 PR tree-optimization/82078
16974 * tree-sra.c (sort_and_splice_var_accesses): Move call to
16975 add_access_to_work_queue...
16976 (build_accesses_from_assign): ...here.
16977 (propagate_all_subaccesses): Make sure racc is the group
16978 representative, if there is one.
16980 2017-09-06 Jakub Jelinek <jakub@redhat.com>
16982 PR middle-end/82095
16983 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
16986 2017-09-06 Richard Biener <rguenther@suse.de>
16988 * gimple-ssa-strength-reduction.c
16989 (find_candidates_dom_walker::before_doom_children): Use a
16990 type and not a mode check.
16992 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
16995 * config/arm/predicates.md (arm_general_adddi_operand): Create new
16997 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
16999 2017-09-05 Jeff Law <law@redhat.com>
17001 PR tree-optimization/64910
17002 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
17003 cases where we have 3 or more operands.
17005 2017-09-05 Jakub Jelinek <jakub@redhat.com>
17007 PR middle-end/81768
17008 * omp-low.c (lower_omp_for): Recompute tree invariant if
17009 gimple_omp_for_initial/final is ADDR_EXPR.
17011 PR middle-end/81768
17012 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
17013 into gimple val before gimplification fo the COND_EXPR.
17015 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
17017 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
17018 REGION_COPY argument.
17019 (thread_through_all_blocks): Remove unused argument to
17020 duplicate_thread_path.
17022 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17023 Alan Hayward <alan.hayward@arm.com>
17024 David Sherwood <david.sherwood@arm.com>
17026 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
17027 Take a scalar_mode rather than a machine_mode.
17028 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17029 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
17030 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17031 (aarch64_gen_adjusted_ldpstp): Likewise.
17032 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
17034 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17035 Alan Hayward <alan.hayward@arm.com>
17036 David Sherwood <david.sherwood@arm.com>
17038 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
17039 Take a scalar_int_mode instead of a machine_mode.
17040 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
17041 (aarch64_output_scalar_simd_mov_immediate): Likewise.
17042 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
17043 (aarch64_simd_attr_length_rglist): Delete.
17044 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
17045 a scalar_int_mode instead of a machine_mode.
17046 (aarch64_add_offset): Likewise.
17047 (aarch64_internal_mov_immediate): Likewise
17048 (aarch64_add_constant_internal): Likewise.
17049 (aarch64_add_constant): Likewise.
17050 (aarch64_movw_imm): Likewise.
17051 (aarch64_rtx_arith_op_extract_p): Likewise.
17052 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
17053 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
17054 Remove assert that the mode isn't a vector.
17055 (aarch64_output_scalar_simd_mov_immediate): Likewise.
17056 (aarch64_expand_mov_immediate): Update calls after above changes.
17057 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
17058 (aarch64_and_bitmask_imm): Check for scalar integer modes.
17059 (aarch64_move_imm): Likewise.
17060 (aarch64_can_const_movi_rtx_p): Likewise.
17061 (aarch64_strip_extend): Likewise.
17062 (aarch64_extr_rtx_p): Likewise.
17063 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
17064 a CONST_INT when the mode parameter is VOIDmode.
17065 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
17067 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17069 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
17070 * stor-layout.c (bitwise_mode_for_mode): Likewise.
17071 (bitwise_type_for_mode): Update accordingly.
17073 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17075 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
17076 * stor-layout.c (mode_for_size_tree): Likewise.
17077 (mode_for_array): Update accordingly.
17078 (layout_decl): Likewise.
17079 (compute_record_mode): Likewise. Only set the mode once.
17081 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17083 * target.def (get_mask_mode): Change return type to opt_mode.
17085 * doc/tm.texi: Regenerate.
17086 * targhooks.h (default_get_mask_mode): Return an opt_mode.
17087 * targhooks.c (default_get_mask_mode): Likewise.
17088 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
17089 * optabs-query.c (can_vec_mask_load_store_p): Update use of
17090 targetm.get_mask_mode.
17091 * tree.c (build_truth_vector_type): Likewise.
17093 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17095 * machmode.h (mode_for_vector): Return an opt_mode.
17096 * stor-layout.c (mode_for_vector): Likewise.
17097 (mode_for_int_vector): Update accordingly.
17098 (layout_type): Likewise.
17099 * config/i386/i386.c (emit_memmov): Likewise.
17100 (ix86_expand_set_or_movmem): Likewise.
17101 (ix86_expand_vector_init): Likewise.
17102 (ix86_get_mask_mode): Likewise.
17103 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
17105 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
17106 * expmed.c (extract_bit_field_1): Likewise.
17107 * expr.c (expand_expr_real_2): Likewise.
17108 * optabs-query.c (can_vec_perm_p): Likewise.
17109 (can_vec_mask_load_store_p): Likewise.
17110 * optabs.c (expand_vec_perm): Likewise.
17111 * targhooks.c (default_get_mask_mode): Likewise.
17112 * tree-vect-stmts.c (vectorizable_store): Likewise.
17113 (vectorizable_load): Likewise.
17114 (get_vectype_for_scalar_type_and_size): Likewise.
17116 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17118 * machmode.h (mode_for_int_vector): New function.
17119 * stor-layout.c (mode_for_int_vector): Likewise.
17120 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
17121 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
17122 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
17123 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
17124 (s390_expand_vcond): Likewise.
17126 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17128 * machmode.h (opt_machine_mode): New type.
17129 (opt_mode<T>): Allow construction from anything that can be
17131 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
17132 (mode_for_size): Return an opt_machine_mode.
17133 * stor-layout.c (mode_for_size): Likewise.
17134 (mode_for_size_tree): Update call accordingly.
17135 (bitwise_mode_for_mode): Likewise.
17136 (make_fract_type): Likewise.
17137 (make_accum_type): Likewise.
17138 * caller-save.c (replace_reg_with_saved_mem): Update call
17140 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17141 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17142 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17143 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17144 * expmed.c (extract_bit_field_1): Likewise.
17145 * reload.c (get_secondary_mem): Likewise.
17146 * varasm.c (assemble_integer): Likewise.
17147 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
17150 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17152 * machmode.h (decimal_float_mode_for_size): New function.
17153 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
17154 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
17155 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
17156 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
17157 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
17158 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
17160 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17162 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
17163 (get_builtin_sync_mode): Likewise.
17164 (expand_ifn_atomic_compare_exchange): Likewise.
17165 (expand_builtin_atomic_clear): Likewise.
17166 (expand_builtin_atomic_test_and_set): Likewise.
17167 (fold_builtin_atomic_always_lock_free): Likewise.
17168 * calls.c (compute_argument_addresses): Likewise.
17169 (emit_library_call_value_1): Likewise.
17170 (store_one_arg): Likewise.
17171 * combine.c (combine_instructions): Likewise.
17172 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
17173 * config/arm/arm.c (arm_function_value): Likewise.
17174 (aapcs_allocate_return_reg): Likewise.
17175 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
17176 * config/i386/i386.c (construct_container): Likewise.
17177 (ix86_gimplify_va_arg): Likewise.
17178 (ix86_expand_sse_cmp): Likewise.
17179 (emit_memmov): Likewise.
17180 (emit_memset): Likewise.
17181 (expand_small_movmem_or_setmem): Likewise.
17182 (ix86_expand_pextr): Likewise.
17183 (ix86_expand_pinsr): Likewise.
17184 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
17185 * config/microblaze/microblaze.c (microblaze_block_move_straight):
17187 * config/mips/mips.c (mips_function_value_1) Likewise.
17188 (mips_block_move_straight): Likewise.
17189 (mips_expand_ins_as_unaligned_store): Likewise.
17190 * config/powerpcspe/powerpcspe.c
17191 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17192 (rs6000_darwin64_record_arg_flush): Likewise.
17193 * config/rs6000/rs6000.c
17194 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17195 (rs6000_darwin64_record_arg_flush): Likewise.
17196 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
17197 (sparc_function_value_1): Likewise.
17198 * config/spu/spu.c (adjust_operand): Likewise.
17199 (spu_emit_branch_or_set): Likewise.
17200 (arith_immediate_p): Likewise.
17201 * emit-rtl.c (gen_lowpart_common): Likewise.
17202 * expr.c (expand_expr_real_1): Likewise.
17203 * function.c (assign_parm_setup_block): Likewise.
17204 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
17205 * reload1.c (alter_reg): Likewise.
17206 * stor-layout.c (mode_for_vector): Likewise.
17207 (layout_type): Likewise.
17209 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17211 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
17212 (spu_convert_move): Likewise.
17213 * lower-subreg.c (resolve_simple_move): Likewise.
17215 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17218 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
17219 define_insn to a define_expand.
17220 (altivec_vsum2sws_direct): New define_insn.
17221 (altivec_vsumsws): Convert from a define_insn to a define_expand.
17223 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17225 * config/arm/arm.c (arm_option_params_internal): Improve setting of
17228 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
17233 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
17234 realign stack if stack alignment needed is less than incoming
17237 2017-09-05 Marek Polacek <polacek@redhat.com>
17240 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
17243 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17245 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
17247 2017-09-05 Richard Biener <rguenther@suse.de>
17249 PR tree-optimization/82084
17250 * fold-const.c (can_native_encode_string_p): Handle wide characters.
17252 2017-09-05 Richard Biener <rguenther@suse.de>
17254 PR tree-optimization/82102
17255 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
17257 2017-09-05 Martin Liska <mliska@suse.cz>
17259 PR tree-optimization/82032
17260 * tree-cfg.c (generate_range_test): New function.
17261 * tree-cfg.h (generate_range_test): Declared here.
17262 * tree-cfgcleanup.c (convert_single_case_switch): New function.
17263 (cleanup_control_expr_graph): Use it.
17264 * tree-switch-conversion.c (try_switch_expansion): Remove
17266 (emit_case_nodes): Use generate_range_test.
17268 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17271 * config/i386/i386.md (*<btsc><mode>_mask): Add
17272 TARGET_USE_BT to insn constraint.
17273 (*btr<mode>_mask): Ditto.
17275 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
17277 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
17278 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
17280 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
17283 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
17284 TARGET_NEON and TARGET_IWMMXT.
17285 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
17286 TARGET_NEON and TARGET_IWMMXT.
17287 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
17289 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17291 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
17292 (ix86_rewrite_tls_address): Ditto.
17293 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
17294 (ix86_rewrite_tls_address_1): Ditto.
17295 (ix86_rewrite_tls_address): Ditto.
17296 * config/i386/predicates.md (tls_address_pattern): New predicate.
17297 * config/i386/i386.md (TLS address splitter): New splitter.
17299 2017-09-04 Richard Biener <rguenther@suse.de>
17301 PR tree-optimization/82084
17302 * fold-const.h (can_native_encode_string_p): Declare.
17303 * fold-const.c (can_native_encode_string_p): Factor out from ...
17304 (native_encode_string): ... here.
17305 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
17306 vectorizing stores from constants we later cannot handle.
17308 2017-09-04 Marek Polacek <polacek@redhat.com>
17311 * doc/invoke.texi: Update -Wtautological-compare documentation.
17313 2017-09-04 Jeff Law <law@redhat.com>
17315 PR tree-optimization/64910
17316 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
17317 swap the first and last operand if the last is a constant.
17319 2017-09-04 Marek Polacek <polacek@redhat.com>
17322 * convert.c (do_narrow): When sanitizing signed integer overflows,
17323 bail out for signed types.
17324 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
17326 2017-09-04 Richard Biener <rguenther@suse.de>
17328 PR tree-optimization/82060
17329 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17330 Move devirtualization after stmt folding and before EH/AB/noreturn
17331 cleanup to get the stmt refs canonicalized. Use a bool instead
17332 of gimple_modified_p since that doesn't work for NOPs. Schedule
17333 NOPs generated by folding for removal.
17335 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17336 Alan Hayward <alan.hayward@arm.com>
17337 David Sherwood <david.sherwood@arm.com>
17339 * coretypes.h (pad_direction): New enum.
17340 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
17341 (FUNCTION_ARG_PADDING): Likewise.
17342 * target.def (function_arg_padding): New hook.
17343 * targhooks.h (default_function_arg_padding): Declare.
17344 * targhooks.c (default_function_arg_padding): New function.
17345 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
17346 (TARGET_FUNCTION_ARG_PADDING): ...this.
17347 * doc/tm.texi: Regenerate.
17348 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
17349 instead of direction.
17350 (compute_argument_addresses): Likewise.
17351 (load_register_parameters): Likewise.
17352 (emit_library_call_value_1): Likewise.
17353 (store_one_arg): Use targetm.calls.function_arg_padding instead
17354 of FUNCTION_ARG_PADDING.
17355 (must_pass_in_stack_var_size_or_pad): Likewise.
17356 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
17357 (emit_group_store): Likewise.
17358 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
17359 instead of FUNCTION_ARG_PADDING.
17360 (emit_push_insn): Likewise, and propagate enum change throughout
17362 * function.h (direction): Delete.
17363 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
17365 * function.c (assign_parm_find_stack_rtl): Likewise.
17366 (assign_parm_setup_block_p): Likewise.
17367 (assign_parm_setup_block): Likewise.
17368 (gimplify_parameters): Likewise.
17369 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
17370 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
17372 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
17373 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17374 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
17375 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
17376 (aarch64_function_arg_padding): ...this new function.
17377 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
17378 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17379 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
17380 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17381 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
17382 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17383 (arm_pad_arg_upward): Replace with...
17384 (arm_function_arg_padding): ...this new function.
17385 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
17387 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
17388 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
17389 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17390 (ia64_hpux_function_arg_padding): Replace with...
17391 (ia64_function_arg_padding): ...this new function. Use pad_direction
17392 instead of direction. Check for TARGET_HPUX.
17393 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
17394 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17395 (iq2000_function_arg_padding): New function.
17396 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
17397 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
17398 (mips_function_arg_padding): ...this new function.
17399 (mips_pad_reg_upward): Update accordingly.
17400 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17401 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
17402 targetm.calls.function_arg_padding.
17403 (FUNCTION_ARG_PADDING): Delete.
17404 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17405 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
17406 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17407 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
17408 (nios2_block_reg_padding): Return pad_direction instead of direction.
17409 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
17410 instead of direction.
17411 (nios2_function_arg_padding): Likewise. Make static.
17412 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17413 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
17414 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
17415 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
17416 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17417 (pa_function_arg_padding): Make static. Return pad_direction instead
17419 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
17420 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17421 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
17422 instead of direction. Use targetm.calls.function_arg_padding.
17423 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
17424 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17425 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
17426 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
17427 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
17429 (function_arg_padding): Rename to...
17430 (rs6000_function_arg_padding): ...this. Make static. Return
17431 pad_direction instead of direction.
17432 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17433 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
17434 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17435 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
17436 instead of direction. Use targetm.calls.function_arg_padding.
17437 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
17438 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17439 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
17440 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
17441 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17442 (function_arg_padding): Rename to...
17443 (rs6000_function_arg_padding): ...this. Make static. Return
17444 pad_direction instead of direction.
17445 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17446 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
17447 * config/s390/s390.c (s390_function_arg_padding): New function.
17448 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17449 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
17450 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
17451 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17452 (function_arg_padding): Rename to...
17453 (sparc_function_arg_padding): ...this. Make static. Return
17454 pad_direction instead of direction.
17455 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
17456 * config/spu/spu.c (spu_function_arg_padding): New function.
17457 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17458 * system.h (FUNCTION_ARG_PADDING): Poison.
17460 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17461 Alan Hayward <alan.hayward@arm.com>
17462 David Sherwood <david.sherwood@arm.com>
17464 * target.def (modes_tieable_p): New hook.
17465 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
17466 (TARGET_MODES_TIEABLE_P): ...this.
17467 * doc/tm.texi.in: Regenerate.
17468 * hooks.h (hook_bool_mode_mode_true): Declare.
17469 * hooks.c (hook_bool_mode_mode_true): New function.
17470 * combine.c (subst): Use targetm.modes_tieable_p instead of
17472 * dse.c (find_shift_sequence): Likewise.
17473 * expmed.c (extract_low_bits): Likewise.
17474 * lower-subreg.c: Include target.h.
17475 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
17477 * rtlanal.c (rtx_cost): Likewise.
17478 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
17479 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
17480 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
17481 (TARGET_MODES_TIEABLE_P): Redefine.
17482 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
17483 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
17484 (TARGET_MODES_TIEABLE_P): Redefine.
17485 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
17486 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
17487 (arc_modes_tieable_p): New function.
17488 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
17489 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
17490 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
17491 (arm_modes_tieable_p): Make static.
17492 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
17493 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
17494 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
17495 (TARGET_MODES_TIEABLE_P): Redefine.
17496 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
17497 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
17498 (TARGET_MODES_TIEABLE_P): Redefine.
17499 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
17500 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
17501 (cr16_modes_tieable_p): New function.
17502 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
17503 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
17504 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
17505 (TRULY_NOOP_TRUNCATION): Update comment.
17506 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
17507 (TRULY_NOOP_TRUNCATION): Update comment.
17508 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
17509 (frv_modes_tieable_p): New function.
17510 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
17511 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
17512 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
17513 (TARGET_MODES_TIEABLE_P): Redefine.
17514 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
17515 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
17516 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
17517 (TARGET_MODES_TIEABLE_P): Redefine.
17518 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
17519 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
17520 (ia64_modes_tieable_p): New function.
17521 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
17522 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
17523 (iq2000_modes_tieable_p): New function.
17524 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
17525 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
17526 (lm32_modes_tieable_p): New function.
17527 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
17528 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
17529 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
17530 (TARGET_MODES_TIEABLE_P): Redefine.
17531 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
17532 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
17533 (m32r_modes_tieable_p): New function.
17534 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
17535 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
17536 (m68k_modes_tieable_p): New function.
17537 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
17538 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
17539 (mcore_modes_tieable_p): New function.
17540 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
17541 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
17543 (TARGET_MODES_TIEABLE_P): Redefine.
17544 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
17545 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
17546 * config/mips/mips.c (mips_modes_tieable_p): Make static.
17547 (TARGET_MODES_TIEABLE_P): Redefine.
17548 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
17549 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
17550 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
17551 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
17552 (mn10300_modes_tieable_p): ...this and make static.
17553 (TARGET_MODES_TIEABLE_P): Redefine.
17554 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
17555 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
17556 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
17557 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
17558 (msp430_modes_tieable_p): Make static.
17559 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
17560 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
17561 (TARGET_MODES_TIEABLE_P): Redefine.
17562 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
17563 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
17564 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
17565 (TARGET_MODES_TIEABLE_P): Redefine.
17566 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
17567 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
17568 * config/pa/pa.c (pa_modes_tieable_p): Make static.
17569 (TARGET_MODES_TIEABLE_P): Redefine.
17570 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
17571 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
17572 (pdp11_modes_tieable_p): New function.
17573 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
17574 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
17575 (rs6000_modes_tieable_p): New function.
17576 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17577 * config/powerpcspe/powerpcspe.md: Update comment.
17578 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
17579 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
17580 (TARGET_MODES_TIEABLE_P): Redefine.
17581 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
17582 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
17583 (rl78_modes_tieable_p): New function.
17584 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
17585 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
17586 (rs6000_modes_tieable_p): New function.
17587 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17588 * config/rs6000/rs6000.md: Update comment.
17589 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
17590 * config/rx/rx.c (rx_modes_tieable_p): New function.
17591 (TARGET_MODES_TIEABLE_P): Redefine.
17592 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
17593 * config/s390/s390.c (s390_modes_tieable_p): New function.
17594 (TARGET_MODES_TIEABLE_P): Redefine.
17595 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
17596 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
17597 (sh_modes_tieable_p): New function.
17598 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
17599 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
17600 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
17601 (sparc_modes_tieable_p): Make static.
17602 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
17603 * config/spu/spu.c (spu_modes_tieable_p): New function.
17604 (TARGET_MODES_TIEABLE_P): Redefine.
17605 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
17606 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
17607 (TARGET_MODES_TIEABLE_P): Redefine.
17608 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
17609 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
17610 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
17611 * config/v850/v850.c (v850_modes_tieable_p): New function.
17612 (TARGET_MODES_TIEABLE_P): Redefine.
17613 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
17614 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
17615 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
17616 (visium_modes_tieable_p): New function.
17617 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
17618 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
17619 (xtensa_modes_tieable_p): New function.
17620 * system.h (MODES_TIEABLE_P): Poison.
17622 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17623 Alan Hayward <alan.hayward@arm.com>
17624 David Sherwood <david.sherwood@arm.com>
17626 * target.def (hard_regno_mode_ok): New hook.
17627 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
17628 (TARGET_HARD_REGNO_MODE_OK): ...this.
17629 * doc/tm.texi.in: Regenerate.
17630 * hooks.h (hook_bool_uint_mode_true): Declare.
17631 * hooks.c (hook_bool_uint_mode_true): New function.
17632 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
17633 HARD_REGNO_MODE_OK.
17634 * genpreds.c (write_insn_preds_c): Add an include of target.h.
17635 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
17636 instead of HARD_REGNO_MODE_OK.
17637 * caller-save.c: Include target.h.
17638 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
17639 HARD_REGNO_MODE_OK.
17640 * combine.c (can_combine_p): Likewise.
17641 (combinable_i3pat): Likewise.
17642 (can_change_dest_mode): Likewise.
17643 * expr.c (init_expr_target): Likewise.
17644 (convert_move): Likewise.
17645 (convert_modes): Likewise.
17646 * ira.c (setup_prohibited_class_mode_regs): Likewise.
17647 (setup_prohibited_mode_move_regs): Likewise.
17648 * ira.h (target_ira): Likewise.
17649 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17650 * lra-constraints.c (process_alt_operands): Likewise.
17651 (split_reg): Likewise.
17652 * recog.c (peep2_find_free_register): Likewise.
17653 * ree.c (combine_reaching_defs): Likewise.
17654 * regcprop.c (maybe_mode_change): Likewise.
17655 * reginfo.c (init_reg_sets_1): Likewise.
17656 (choose_hard_reg_mode): Likewise.
17657 (simplifiable_subregs): Likewise.
17658 * regrename.c (check_new_reg_p): Likewise.
17659 * reload.c (find_valid_class): Likewise.
17660 (find_valid_class_1): Likewise.
17661 (reload_inner_reg_of_subreg): Likewise.
17662 (push_reload): Likewise.
17663 (combine_reloads): Likewise.
17664 (find_dummy_reload): Likewise.
17665 (find_reloads): Likewise.
17666 * reload1.c (find_reg): Likewise.
17667 (set_reload_reg): Likewise.
17668 (allocate_reload_reg): Likewise.
17669 (choose_reload_regs): Likewise.
17670 (reload_adjust_reg_for_temp): Likewise.
17671 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
17672 (simplify_subreg_regno): Likewise.
17673 * sel-sched.c (init_regs_for_mode): Likewise.
17674 * varasm.c (make_decl_rtl): Likewise.
17675 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
17676 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
17677 HARD_REGNO_MODE_OK.
17678 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
17679 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
17680 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17681 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
17682 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
17683 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17684 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
17685 (arc_mode_class): Delete.
17686 (HARD_REGNO_MODE_OK): Delete.
17687 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17688 (arc_hard_regno_mode_ok): Rename old array to...
17689 (arc_hard_regno_mode_ok_modes): ...this.
17690 (arc_conditional_register_usage): Update accordingly.
17691 (arc_mode_class): Make static.
17692 (arc_hard_regno_mode_ok): New function.
17693 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
17694 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
17695 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17696 (arm_hard_regno_mode_ok): Make static.
17697 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
17698 HARD_REGNO_MODE_OK.
17699 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
17700 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
17701 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
17703 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17704 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
17705 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
17706 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
17707 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
17708 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17709 * config/bfin/predicates.md (valid_reg_operand): Use
17710 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
17711 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
17712 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
17713 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17714 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
17715 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
17716 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17717 (cr16_hard_regno_mode_ok): Make static and return a bool.
17718 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
17719 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17720 (cris_hard_regno_mode_ok): New function.
17721 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
17722 (epiphany_mode_class): Delete.
17723 (HARD_REGNO_MODE_OK): Delete.
17724 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
17725 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17726 (hard_regno_mode_ok): Rename to...
17727 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
17728 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
17729 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
17730 HARD_REGNO_MODE_OK.
17731 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
17732 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
17733 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17734 (frv_hard_regno_mode_ok): Make static and return a bool.
17735 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
17736 HARD_REGNO_MODE_OK.
17737 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
17738 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
17739 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
17740 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
17742 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17743 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
17744 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
17745 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
17747 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17748 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
17749 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17750 (ia64_hard_regno_mode_ok): New function.
17751 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
17752 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17753 (iq2000_hard_regno_mode_ok): New function.
17754 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
17755 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17756 (lm32_hard_regno_mode_ok): New function.
17757 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
17758 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
17759 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
17760 instead of HARD_REGNO_MODE_OK.
17761 (m32c_hard_regno_ok): Rename to...
17762 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
17763 (m32c_cannot_change_mode_class): Update accordingly.
17764 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17765 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
17766 (m32r_mode_class): Delete.
17767 (HARD_REGNO_MODE_OK): Delete.
17768 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17769 (m32r_hard_regno_mode_ok): Rename to...
17770 (m32r_hard_regno_modes): ...this.
17771 (m32r_mode_class): Make static.
17772 (m32r_hard_regno_mode_ok): New function.
17773 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
17774 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
17775 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17776 (m68k_hard_regno_mode_ok): Make static.
17777 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
17778 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17779 (mcore_hard_regno_mode_ok): New function.
17780 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
17781 (HARD_REGNO_MODE_OK): Delete.
17782 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
17784 (microblaze_hard_regno_mode_ok_p): ...this and make static.
17785 (microblaze_hard_regno_mode_ok): New function.
17786 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17787 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
17788 (mips_hard_regno_mode_ok): Delete.
17789 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
17790 (mips_hard_regno_mode_ok_p): ...this and make static.
17791 (mips_hard_regno_mode_ok_p): Rename to...
17792 (mips_hard_regno_mode_ok_uncached): ...this.
17793 (mips_hard_regno_mode_ok): New function.
17794 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
17795 of HARD_REGNO_MODE_OK.
17796 (mips_option_override): Update after above name changes.
17797 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17798 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
17799 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
17800 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
17801 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
17802 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17803 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
17804 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
17805 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17806 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17807 (msp430_hard_regno_mode_ok): Make static and return a bool.
17808 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17809 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17810 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17812 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17813 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17814 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17815 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17816 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17817 (PA_HARD_REGNO_MODE_OK): ...this
17818 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17819 (PA_HARD_REGNO_MODE_OK): ...this.
17820 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17821 (pa_hard_regno_mode_ok): New function.
17822 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17823 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17824 (pdp11_hard_regno_mode_ok): New function.
17825 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17826 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17828 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17830 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17831 (rs6000_hard_regno_mode_ok): Rename to...
17832 (rs6000_hard_regno_mode_ok_uncached): ...this.
17833 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17834 (rs6000_hard_regno_mode_ok): New function.
17835 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17836 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17837 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17838 (riscv_hard_regno_mode_ok): ...this and make static.
17839 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17840 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17841 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17842 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17843 (rl78_hard_regno_mode_ok): Make static and return bool.
17844 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
17845 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17847 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17848 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17849 (rs6000_hard_regno_mode_ok): Rename to...
17850 (rs6000_hard_regno_mode_ok_uncached): ...this.
17851 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17852 (rs6000_hard_regno_mode_ok): New function.
17853 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17854 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17855 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17856 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17857 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17858 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17859 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17860 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17861 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17862 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17863 (sh_hard_regno_mode_ok): Make static.
17864 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17865 instead of HARD_REGNO_MODE_OK.
17866 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17867 (sparc_mode_class): Delete.
17868 (HARD_REGNO_MODE_OK): Delete.
17869 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17870 (hard_regno_mode_classes): Make static.
17871 (sparc_mode_class): Likewise.
17872 (sparc_hard_regno_mode_ok): New function.
17873 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17874 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17875 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17877 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17878 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17879 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17880 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17881 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17882 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17883 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17884 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17885 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17886 (visium_hard_regno_mode_ok): New function.
17887 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17888 instead of HARD_REGNO_MODE_OK.
17889 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17890 (HARD_REGNO_MODE_OK): Delete.
17891 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17892 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17893 (xtensa_option_override): Update accordingly.
17894 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17895 (xtensa_hard_regno_mode_ok): New function.
17896 * system.h (HARD_REGNO_MODE_OK): Poison.
17898 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17899 Alan Hayward <alan.hayward@arm.com>
17900 David Sherwood <david.sherwood@arm.com>
17902 * target.def (hard_regno_call_part_clobbered): New hook.
17903 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17904 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17905 * doc/tm.texi: Regenerate.
17906 * hooks.h (hook_bool_uint_mode_false): Declare.
17907 * hooks.c (hook_bool_uint_mode_false): New function.
17908 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17909 * cselib.c (cselib_process_insn): Use
17910 targetm.hard_regno_call_part_clobbered instead of
17911 HARD_REGNO_CALL_PART_CLOBBERED.
17912 * ira-conflicts.c (ira_build_conflicts): Likewise.
17913 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17914 * lra-constraints.c (need_for_call_save_p): Likewise.
17915 * lra-lives.c: Include target.h.
17916 (check_pseudos_live_through_calls): Use
17917 targetm.hard_regno_call_part_clobbered instead of
17918 HARD_REGNO_CALL_PART_CLOBBERED.
17919 * regcprop.c: Include target.h.
17920 (copyprop_hardreg_forward_1): Use
17921 targetm.hard_regno_call_part_clobbered instead of
17922 HARD_REGNO_CALL_PART_CLOBBERED.
17923 * reginfo.c (choose_hard_reg_mode): Likewise.
17924 * regrename.c (check_new_reg_p): Likewise.
17925 * reload.c (find_equiv_reg): Likewise.
17926 * reload1.c (emit_reload_insns): Likewise.
17927 * sched-deps.c (deps_analyze_insn): Likewise.
17928 * sel-sched.c (init_regs_for_mode): Likewise.
17929 (mark_unavailable_hard_regs): Likewise.
17930 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
17931 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17932 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
17934 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17935 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17936 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
17938 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
17940 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17941 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17942 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
17944 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17945 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17946 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
17948 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17949 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
17951 * config/powerpcspe/powerpcspe.c
17952 (rs6000_hard_regno_call_part_clobbered): New function.
17953 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17954 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17955 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
17957 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17958 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17959 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
17961 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17962 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17963 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
17965 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17966 Alan Hayward <alan.hayward@arm.com>
17967 David Sherwood <david.sherwood@arm.com>
17969 * rtl.h (subreg_memory_offset): Declare.
17970 * emit-rtl.c (subreg_memory_offset): New function.
17971 * expmed.c (store_bit_field_1): Use it.
17972 * expr.c (undefined_operand_subword_p): Likewise.
17973 * simplify-rtx.c (simplify_subreg): Likewise.
17975 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
17977 PR rtl-optimization/57448
17980 * optabs.c (expand_atomic_load): Place compiler memory barriers if
17981 using atomic_load pattern.
17982 (expand_atomic_store): Likewise.
17984 2017-09-04 Jakub Jelinek <jakub@redhat.com>
17987 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
17988 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
17989 handling. Use replace_call_with_value with NULL instead of
17990 gsi_replace, unlink_stmt_vdef and release_defs.
17992 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
17995 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
17997 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
18000 * rtl.h (emit_library_call_value_1): Declare.
18001 (emit_library_call): Replace declaration with a series of overloads.
18002 Remove the parameter count argument.
18003 (emit_library_call_value): Likewise.
18004 * calls.c (emit_library_call_value_1): Make global. Replace varargs
18005 with an "rtx_mode_t *".
18006 (emit_library_call_value): Delete.
18007 (emit_library_call): Likewise.
18008 * asan.c (asan_emit_stack_protection): Update calls accordingly.
18009 (asan_emit_allocas_unpoison): Likewise.
18010 * builtins.c (expand_builtin_powi): Likewise.
18011 (expand_asan_emit_allocas_unpoison): Likewise.
18012 * cfgexpand.c (expand_main_function): Likewise.
18013 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
18014 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
18015 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
18016 * config/arm/arm.c (arm_trampoline_init): Likewise.
18017 (arm_call_tls_get_addr): Likewise.
18018 (arm_expand_divmod_libfunc): Likewise.
18019 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
18020 (smulsi3_highpart): Likewise.
18021 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
18022 (c6x_expand_compare): Likewise.
18023 (c6x_expand_movmem): Likewise.
18024 * config/frv/frv.c (frv_trampoline_init): Likewise.
18025 * config/i386/i386.c (ix86_trampoline_init): Likewise.
18026 (ix86_expand_divmod_libfunc): Likewise.
18027 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
18028 (ia64_expand_compare): Likewise.
18029 (ia64_profile_hook): Likewise.
18030 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
18031 (nonlocal_goto): Likewise.
18032 (restore_stack_nonlocal): Likewise.
18033 * config/m32r/m32r.c (block_move_call): Likewise.
18034 (m32r_trampoline_init): Likewise.
18035 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
18036 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
18037 (m68k_call_m68k_read_tp): Likewise.
18038 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
18039 (microblaze_expand_divide): Likewise.
18040 * config/mips/mips.h (mips_args): Likewise.
18041 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
18042 (MIPS_ICACHE_SYNC): Likewise.
18043 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
18044 (nios2_trampoline_init): Likewise.
18045 * config/pa/pa.c (hppa_tls_call): Likewise.
18046 (pa_trampoline_init): Likewise.
18047 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
18048 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
18049 (expand_strn_compare): Likewise.
18050 (rs6000_generate_compare): Likewise.
18051 (rs6000_expand_float128_convert): Likewise.
18052 (output_profile_hook): Likewise.
18053 (rs6000_trampoline_init): Likewise.
18054 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
18055 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
18056 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
18057 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
18058 (rs6000_generate_compare): Likewise.
18059 (rs6000_expand_float128_convert): Likewise.
18060 (output_profile_hook): Likewise.
18061 (rs6000_trampoline_init): Likewise.
18062 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
18063 * config/sh/sh.c (sh_trampoline_init): Likewise.
18064 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
18065 (sparc_emit_float_lib_cmp): Likewise.
18066 (sparc32_initialize_trampoline): Likewise.
18067 (sparc64_initialize_trampoline): Likewise.
18068 (sparc_profile_hook): Likewise.
18069 * config/spu/spu.c (ea_load_store): Likewise.
18070 * config/spu/spu.md (floatunssidf2): Likewise.
18071 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
18072 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
18073 * config/visium/visium.c (expand_block_move_4): Likewise.
18074 (expand_block_move_2): Likewise.
18075 (expand_block_move_1): Likewise.
18076 (expand_block_set_4): Likewise.
18077 (expand_block_set_2): Likewise.
18078 (expand_block_set_1): Likewise.
18079 (visium_trampoline_init): Likewise.
18080 (visium_profile_hook): Likewise.
18081 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
18082 (xtensa_setup_frame_addresses): Likewise.
18083 (xtensa_trampoline_init): Likewise.
18084 * except.c (sjlj_emit_function_enter): Likewise.
18085 (sjlj_emit_function_exit): Likewise.
18086 * explow.c (allocate_dynamic_stack_space): Likewise.
18087 (probe_stack_range): Likewise.
18088 * expr.c (convert_mode_scalar): Likewise.
18089 * optabs.c (expand_binop): Likewise.
18090 (expand_twoval_binop_libfunc): Likewise.
18091 (expand_unop): Likewise.
18092 (prepare_cmp_insn): Likewise.
18093 (prepare_float_lib_cmp): Likewise.
18094 (expand_float): Likewise.
18095 (expand_fix): Likewise.
18096 (expand_fixed_convert): Likewise.
18097 (maybe_emit_sync_lock_test_and_set): Likewise.
18098 (expand_atomic_compare_and_swap): Likewise.
18099 (expand_mem_thread_fence): Likewise.
18100 (expand_atomic_fetch_op): Likewise.
18102 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
18104 * doc/generic.texi (OpenACC): Adjust URL.
18105 * doc/invoke.texi (C Dialect Options): Ditto.
18107 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
18109 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
18110 predicate for operand 1. Add (m,<S>) constraint.
18111 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
18112 Prevent memory operand 1 with register operand 2.
18114 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
18116 PR rtl-optimization/82024
18117 * combine.c (try_combine): If the combination result is a PARALLEL,
18118 and we only need to retain the SET in there that would be placed
18119 at I2, check that we can place that at I3 instead, before doing so.
18121 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18124 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
18126 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
18127 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
18130 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
18131 Jeff Law <law@redhat.com>
18133 * varasm.c (bss_initializer_p): Do not put constants into .bss
18134 (categorize_decl_for_section): Handle bss_initializer_p returning
18135 false when DECL_INITIAL is NULL.
18137 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18140 * config/s390/s390.c (s390_can_inline_p): New function.
18142 2017-09-01 Jeff Law <law@redhat.com>
18144 PR tree-optimization/82052
18145 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18146 Always initialize the returned slot after a hash table miss
18147 when INSERT is true.
18149 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
18151 * config/s390/s390.md (mem_signal_fence): Remove.
18152 * doc/md.texi (mem_signal_fence): Remove.
18153 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
18155 * target-insns.def (mem_signal_fence): Remove.
18157 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18160 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
18163 * asan.c (create_odr_indicator): Strip name encoding from assembler
18164 name before appending it after __odr_asan_.
18166 2017-09-01 Martin Liska <mliska@suse.cz>
18168 PR tree-optimization/82059
18169 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
18170 frequency only when an edge is redirected.
18172 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18174 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
18175 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
18176 (arc_conditional_register_usage): Remove ARC600 lp_count
18178 (arc_file_start): Emit Tag_ARC_CPU_variation.
18179 (arc_can_use_doloop_p): New conditions to use ZOLs.
18180 (hwloop_fail): New function.
18181 (hwloop_optimize): Likewise.
18182 (hwloop_pattern_reg): Likewise.
18183 (arc_doloop_hooks): New struct, to be used with reorg_loops.
18184 (arc_reorg_loops): New function, calls reorg_loops.
18185 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
18186 (arc600_corereg_hazard): Remove ZOL checking, case handled by
18188 (arc_loop_hazard): Remove function, functionality moved into
18190 (arc_hazard): Remove arc_loop_hazard call.
18191 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
18192 into hwloop_optimize.
18193 (arc_label_align): Remove ZOL handling.
18194 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
18195 * config/arc/arc.md (doloop_begin): Remove pattern.
18196 (doloop_begin_i): Likewise.
18197 (doloop_end_i): Likewise.
18198 (doloop_fallback): Likewise.
18199 (doloop_fallback_m): Likewise.
18200 (doloop_end): Reimplement expand.
18201 (arc_lp): New pattern for LP instruction.
18202 (loop_end): New pattern.
18203 (loop_fail): Likewise.
18204 (decrement_and_branch_until_zero): Likewise.
18205 * config/arc/arc.opt (mlpc-width): New option.
18206 * doc/invoke.texi (mlpc-width): Document option.
18208 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18210 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
18211 (arc_ccfsm_advance): Fix checking for delay slots.
18212 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
18214 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18216 * config/arc/arc.md (movqi_insn): Add stores to save constant long
18218 (movhi_insn): Update store instruction constraint which are saving
18219 6-bit short immediates.
18220 (movsi_insn): Consider also short scaled load operations.
18221 (zero_extendhisi2_i): Use Usd constraint instead of T.
18222 (extendhisi2_i): Add q constraint.
18223 (arc_clzsi2): Add type and length attributes.
18224 (arc_ctzsi2): Likewise.
18225 * config/arc/constraints.md (Usc): Update constraint, the
18226 assembler can parse two relocations for a single instruction.
18228 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18230 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
18231 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
18233 2017-08-31 Olivier Hainque <hainque@adacore.com>
18235 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
18236 match as powerpc-wrs-vxworks*.
18238 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
18240 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
18241 register constraint for by-element operand.
18242 (aarch64_mls_elt_merge<mode>): Likewise.
18244 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18246 * config/arc/arc.c (arc_can_follow_jump): Check for short
18249 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18251 * config.gcc: Use g.opt for arc.
18252 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
18253 functionality moved to ...
18254 (legitimate_scaled_address_p): New function, ...here.
18255 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
18256 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
18257 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
18259 (arc_override_options): Handle G option.
18260 (arc_output_pic_addr_const): Correct function definition.
18261 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
18262 (arc_decl_anon_ns_mem_p): Delete.
18263 (arc_in_small_data_p): Overhaul this function to take into
18264 consideration the value given via G option.
18265 (arc_rewrite_small_data_1): Renamed and corrected old
18266 arc_rewrite_small_data function.
18267 (arc_rewrite_small_data): New function.
18268 (small_data_pattern): Don't use pic_offset_table_rtx.
18269 * config/arc/arc.h (CC1_SPEC): Recognize G option.
18270 * config/arc/simdext.md (movmisalignv2hi): Use
18271 prepare_move_operands function.
18273 (movmisalign*): Likewise.
18274 * doc/invoke.texi (ARC options): Document -G option.
18276 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18278 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
18280 * config/arc/arc.c (arc_print_operand): Output scalled address for
18281 sdata whenever is possible.
18282 (arc_in_small_data_p): Allow sdata for 64bit datum when double
18283 load/stores are available.
18284 (compact_sda_memory_operand): Check for the alignment required by
18285 code density instructions.
18286 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
18288 * config/arc/constraints.md (Usd): Update constraint.
18289 (Us0): New constraint.
18290 (Usc): Update constraint.
18292 2017-08-31 Richard Biener <rguenther@suse.de>
18294 PR middle-end/82054
18295 * dwarf2out.c (dwarf2out_early_global_decl): Process each
18296 function only once.
18298 2017-08-31 Tamar Christina <tamar.christina@arm.com>
18300 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
18301 Resize type_signature.
18303 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18304 Alan Hayward <alan.hayward@arm.com>
18305 David Sherwood <david.sherwood@arm.com>
18307 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
18308 subregs whose inner modes can be stored in GPRs.
18309 (aarch64_classify_index): Likewise.
18311 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18312 Alan Hayward <alan.hayward@arm.com>
18313 David Sherwood <david.sherwood@arm.com>
18315 * config/aarch64/iterators.md (V_cmp_result): Rename to...
18316 (V_INT_EQUIV): ...this.
18317 (v_cmp_result): Rename to...
18318 (v_int_equiv): ...this.
18319 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
18320 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
18321 (copysign<mode>3): Likewise.
18322 (aarch64_simd_bsl<mode>_internal): Likewise.
18323 (aarch64_simd_bsl<mode>): Likewise.
18324 (vec_cmp<mode><mode>): Likewise.
18325 (vcond<mode><mode>): Likewise.
18326 (vcond<v_cmp_mixed><mode>): Likewise.
18327 (vcondu<mode><v_cmp_mixed>): Likewise.
18328 (aarch64_cm<optab><mode>): Likewise.
18329 (aarch64_cmtst<mode>): Likewise.
18330 (aarch64_fac<optab><mode>): Likewise.
18331 (vec_perm_const<mode>): Likewise.
18332 (vcond_mask_<mode><v_cmp_result>): Rename to...
18333 (vcond_mask_<mode><v_int_equiv>): ...this.
18334 (vec_cmp<mode><v_cmp_result>): Rename to...
18335 (vec_cmp<mode><v_int_equiv>): ...this.
18337 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18338 Alan Hayward <alan.hayward@arm.com>
18339 David Sherwood <david.sherwood@arm.com>
18341 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
18343 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
18344 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
18345 (UNSPEC_LD4_DREG): New unspecs.
18346 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
18347 (aarch64_ld2<mode>_dreg_be): Replace with...
18348 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
18350 (aarch64_ld3<mode>_dreg_le)
18351 (aarch64_ld3<mode>_dreg_be): Replace with...
18352 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
18354 (aarch64_ld4<mode>_dreg_le)
18355 (aarch64_ld4<mode>_dreg_be): Replace with...
18356 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
18359 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18361 PR tree-optimization/81987
18362 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
18363 insert an initializer in a location not dominated by the stride
18366 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18368 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
18369 on the entire header of the finally block in the fallthru case.
18371 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18373 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
18375 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
18377 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
18378 rs6000_emit_move_from_cr and call renamed function.
18379 (rs6000_emit_prologue): Call renamed functions.
18380 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
18381 movesi_from_cr, remove volatile CRs.
18383 2017-08-30 Jon Beniston <jon@beniston.com>
18384 Richard Biener <rguenther@suse.de>
18386 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
18387 of VECTOR_MODE_P check.
18388 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
18389 element vector types.
18391 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18393 * df.h (df_read_modify_subreg_p): Remove in favor of...
18394 * rtl.h (read_modify_subreg_p): ...this new function. Take a
18395 const_rtx instead of an rtx.
18396 * cprop.c (local_cprop_find_used_regs): Update accordingly.
18397 * df-problems.c (df_word_lr_mark_ref): Likewise.
18398 * ira-lives.c (mark_pseudo_reg_live): Likewise.
18399 (mark_pseudo_reg_dead): Likewise.
18400 (mark_ref_dead): Likewise.
18401 * reginfo.c (init_subregs_of_mode): Likewise.
18402 * sched-deps.c (sched_analyze_1): Likewise.
18403 * df-scan.c (df_def_record_1): Likewise.
18404 (df_uses_record): Likewise.
18405 (df_read_modify_subreg_p): Remove in favor of...
18406 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
18407 const_rtx instead of an rtx.
18409 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18410 Alan Hayward <alan.hayward@arm.com>
18411 David Sherwood <david.sherwood@arm.com>
18413 * rtl.h (partial_subreg_p): New function.
18414 * caller-save.c (save_call_clobbered_regs): Use it.
18415 * calls.c (expand_call): Likewise.
18416 * combine.c (combinable_i3pat): Likewise.
18417 (simplify_set): Likewise.
18418 (make_extraction): Likewise.
18419 (make_compound_operation_int): Likewise.
18420 (gen_lowpart_or_truncate): Likewise.
18421 (force_to_mode): Likewise.
18422 (make_field_assignment): Likewise.
18423 (reg_truncated_to_mode): Likewise.
18424 (record_truncated_value): Likewise.
18425 (move_deaths): Likewise.
18426 * cse.c (record_jump_cond): Likewise.
18427 (cse_insn): Likewise.
18428 * cselib.c (cselib_lookup_1): Likewise.
18429 * expmed.c (extract_bit_field_using_extv): Likewise.
18430 * function.c (assign_parm_setup_reg): Likewise.
18431 * ifcvt.c (noce_convert_multiple_sets): Likewise.
18432 * ira-build.c (create_insn_allocnos): Likewise.
18433 * lra-coalesce.c (merge_pseudos): Likewise.
18434 * lra-constraints.c (match_reload): Likewise.
18435 (simplify_operand_subreg): Likewise.
18436 (curr_insn_transform): Likewise.
18437 * lra-lives.c (process_bb_lives): Likewise.
18438 * lra.c (new_insn_reg): Likewise.
18439 (lra_substitute_pseudo): Likewise.
18440 * regcprop.c (mode_change_ok): Likewise.
18441 (maybe_mode_change): Likewise.
18442 (copyprop_hardreg_forward_1): Likewise.
18443 * reload.c (push_reload): Likewise.
18444 (find_reloads): Likewise.
18445 (find_reloads_subreg_address): Likewise.
18446 * reload1.c (alter_reg): Likewise.
18447 (eliminate_regs_1): Likewise.
18448 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18450 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
18452 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
18453 back to if statements, including unpack.
18455 2017-08-30 Martin Liska <mliska@suse.cz>
18457 PR inline-asm/82001
18458 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
18460 (func_checker::compare_asm_inputs_outputs): ... this function.
18461 (func_checker::compare_gimple_asm): Use the function to compare
18462 also ASM constrains.
18463 * ipa-icf-gimple.h: Rename the function.
18465 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18466 Alan Hayward <alan.hayward@arm.com>
18467 David Sherwood <david.sherwood@arm.com>
18469 * coretypes.h (complex_mode): New type.
18470 * gdbhooks.py (build_pretty_printer): Handle it.
18471 * machmode.h (complex_mode): New class.
18472 (complex_mode::includes_p): New function.
18473 (is_complex_int_mode): Likewise.
18474 (is_complex_float_mode): Likewise.
18475 * genmodes.c (get_mode_class): Handle complex mode classes.
18476 * function.c (expand_function_end): Use is_complex_int_mode.
18478 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18479 Alan Hayward <alan.hayward@arm.com>
18480 David Sherwood <david.sherwood@arm.com>
18482 * coretypes.h (scalar_mode_pod): New typedef.
18483 * gdbhooks.py (build_pretty_printer): Handle it.
18484 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
18485 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
18486 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
18487 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
18488 as_a <scalar_mode>.
18490 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18491 Alan Hayward <alan.hayward@arm.com>
18492 David Sherwood <david.sherwood@arm.com>
18494 * machmode.h (mode_for_vector): Take a scalar_mode instead
18496 * stor-layout.c (mode_for_vector): Likewise.
18497 * explow.c (promote_mode): Use as_a <scalar_mode>.
18498 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
18500 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18501 Alan Hayward <alan.hayward@arm.com>
18502 David Sherwood <david.sherwood@arm.com>
18504 * target.def (preferred_simd_mode): Take a scalar_mode
18505 instead of a machine_mode.
18506 * targhooks.h (default_preferred_simd_mode): Likewise.
18507 * targhooks.c (default_preferred_simd_mode): Likewise.
18508 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
18509 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
18510 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
18511 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
18512 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
18513 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
18514 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
18515 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
18517 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
18518 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
18519 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
18520 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
18521 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
18522 * doc/tm.texi: Regenerate.
18523 * optabs-query.c (can_vec_mask_load_store_p): Return false for
18526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18527 Alan Hayward <alan.hayward@arm.com>
18528 David Sherwood <david.sherwood@arm.com>
18530 * target.def (scalar_mode_supported_p): Take a scalar_mode
18531 instead of a machine_mode.
18532 * targhooks.h (default_scalar_mode_supported_p): Likewise.
18533 * targhooks.c (default_scalar_mode_supported_p): Likewise.
18534 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
18535 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18536 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
18537 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
18538 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
18539 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
18540 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
18541 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
18542 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
18543 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
18544 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
18545 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
18547 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
18548 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
18549 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18550 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18551 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
18553 * doc/tm.texi: Regenerate.
18555 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18556 Alan Hayward <alan.hayward@arm.com>
18557 David Sherwood <david.sherwood@arm.com>
18559 * coretypes.h (opt_scalar_mode): New typedef.
18560 * gdbhooks.py (build_pretty_printers): Handle it.
18561 * machmode.h (mode_iterator::get_2xwider): Add overload for
18563 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
18565 * expr.c (convert_mode_scalar): Likewise.
18566 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18567 * optabs.c (expand_float): Likewise.
18568 (expand_fix): Likewise.
18569 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18571 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18572 Alan Hayward <alan.hayward@arm.com>
18573 David Sherwood <david.sherwood@arm.com>
18575 * optabs.c (expand_float): Explicitly check for scalars before
18576 using a branching expansion.
18577 (expand_fix): Likewise.
18579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18580 Alan Hayward <alan.hayward@arm.com>
18581 David Sherwood <david.sherwood@arm.com>
18583 * expr.c (convert_mode): Split scalar handling out into...
18584 (convert_mode_scalar): ...this new function. Treat the modes
18587 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18588 Alan Hayward <alan.hayward@arm.com>
18589 David Sherwood <david.sherwood@arm.com>
18591 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
18593 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
18595 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18596 Alan Hayward <alan.hayward@arm.com>
18597 David Sherwood <david.sherwood@arm.com>
18599 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
18600 rather than a machine_mode.
18601 (fixed_from_string): Likewise.
18602 (fixed_convert): Likewise.
18603 (fixed_convert_from_int): Likewise.
18604 (fixed_convert_from_real): Likewise.
18605 (real_convert_from_fixed): Likewise.
18606 * fixed-value.c (fixed_from_double_int): Likewise.
18607 (fixed_from_string): Likewise.
18608 (fixed_convert): Likewise.
18609 (fixed_convert_from_int): Likewise.
18610 (fixed_convert_from_real): Likewise.
18611 (real_convert_from_fixed): Likewise.
18612 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
18614 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18615 Alan Hayward <alan.hayward@arm.com>
18616 David Sherwood <david.sherwood@arm.com>
18618 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
18619 of separate mode class checks. Do not allow vector modes here.
18620 (immed_wide_int_const): Use as_a <scalar_mode>.
18621 * explow.c (trunc_int_for_mode): Likewise.
18622 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
18623 (wi::shwi): Likewise.
18624 (wi::min_value): Likewise.
18625 (wi::max_value): Likewise.
18626 * dwarf2out.c (loc_descriptor): Likewise.
18627 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
18628 for CONST_WIDE_INT.
18630 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18631 Alan Hayward <alan.hayward@arm.com>
18632 David Sherwood <david.sherwood@arm.com>
18634 * tree.h (SCALAR_TYPE_MODE): New macro.
18635 * expr.c (expand_expr_addr_expr_1): Use it.
18636 (expand_expr_real_2): Likewise.
18637 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
18638 (fold_convert_const_fixed_from_int): Likewise.
18639 (fold_convert_const_fixed_from_real): Likewise.
18640 (native_encode_fixed): Likewise
18641 (native_encode_complex): Likewise
18642 (native_encode_vector): Likewise.
18643 (native_interpret_fixed): Likewise.
18644 (native_interpret_real): Likewise.
18645 (native_interpret_complex): Likewise.
18646 (native_interpret_vector): Likewise.
18647 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
18648 (simd_clone_adjust_argument_types): Likewise.
18649 (simd_clone_init_simd_arrays): Likewise.
18650 (simd_clone_adjust): Likewise.
18651 * stor-layout.c (layout_type): Likewise.
18652 * tree.c (build_minus_one_cst): Likewise.
18653 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18654 * tree-inline.c (estimate_move_cost): Likewise.
18655 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
18656 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
18657 (vectorizable_reduction): Likewise.
18658 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
18659 (vect_recog_mixed_size_cond_pattern): Likewise.
18660 (check_bool_pattern): Likewise.
18661 (adjust_bool_pattern): Likewise.
18662 (search_type_for_mask_1): Likewise.
18663 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
18664 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18665 (vectorizable_load): Likewise.
18666 (vectorizable_store): Likewise.
18667 * ubsan.c (ubsan_encode_value): Likewise.
18668 * varasm.c (output_constant): Likewise.
18670 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18671 Alan Hayward <alan.hayward@arm.com>
18672 David Sherwood <david.sherwood@arm.com>
18674 * coretypes.h (scalar_mode): New class.
18675 * machmode.h (scalar_mode): Likewise.
18676 (scalar_mode::includes_p): New function.
18677 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
18678 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
18679 * genmodes.c (get_mode_class): Handle remaining scalar modes.
18680 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
18681 * expmed.c (store_bit_field_1): Likewise.
18682 (extract_bit_field_1): Likewise.
18683 * expr.c (write_complex_part): Likewise.
18684 (read_complex_part): Likewise.
18685 (emit_move_complex_push): Likewise.
18686 (expand_expr_real_2): Likewise.
18687 * function.c (assign_parm_setup_reg): Likewise.
18688 (assign_parms_unsplit_complex): Likewise.
18689 * optabs.c (expand_binop): Likewise.
18690 * rtlanal.c (subreg_get_info): Likewise.
18691 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18692 * varasm.c (output_constant_pool_2): Likewise.
18694 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18695 Alan Hayward <alan.hayward@arm.com>
18696 David Sherwood <david.sherwood@arm.com>
18698 * expmed.c (extract_high_half): Use scalar_int_mode and remove
18700 (expmed_mult_highpart_optab): Likewise.
18701 (expmed_mult_highpart): Likewise.
18703 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18704 Alan Hayward <alan.hayward@arm.com>
18705 David Sherwood <david.sherwood@arm.com>
18707 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
18708 instead of a machine_mode.
18709 (builtin_memset_read_str): Likewise.
18710 * builtins.c (c_readstr): Likewise.
18711 (builtin_memcpy_read_str): Likewise.
18712 (builtin_strncpy_read_str): Likewise.
18713 (builtin_memset_read_str): Likewise.
18714 (builtin_memset_gen_str): Likewise.
18715 (expand_builtin_signbit): Use scalar_int_mode for local variables.
18716 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
18717 instead of a machine_mode.
18718 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
18720 (make_extraction): Likewise.
18721 (try_widen_shift_mode): Take and return scalar_int_modes instead
18723 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
18724 a scalar_int_mode instead of a machine_mode.
18725 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
18726 (avr_addr_space_pointer_mode): Likewise.
18727 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
18728 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
18729 (msp430_unwind_word_mode): Likewise.
18730 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
18731 (spu_addr_space_pointer_mode): Likewise.
18732 (spu_addr_space_address_mode): Likewise.
18733 (spu_libgcc_cmp_return_mode): Likewise.
18734 (spu_libgcc_shift_count_mode): Likewise.
18735 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
18736 (rl78_addr_space_pointer_mode): Likewise.
18737 (fl78_unwind_word_mode): Likewise.
18738 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
18740 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
18741 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
18742 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
18743 (mips_valid_pointer_mode): Likewise.
18744 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
18745 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
18746 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
18748 (ft32_addr_space_address_mode): Likewise.
18749 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
18750 scalar_int_mode instead of a machine_mode.
18751 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
18753 (m32c_addr_space_address_mode): Likewise.
18754 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
18755 (rs6000_eh_return_filter_mode): Likewise.
18756 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
18757 (rs6000_eh_return_filter_mode): Likewise.
18758 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
18759 (s390_libgcc_shift_count_mode): Likewise.
18760 (s390_unwind_word_mode): Likewise.
18761 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
18763 * target.def (mode_rep_extended): Likewise.
18764 (valid_pointer_mode): Likewise.
18765 (addr_space.valid_pointer_mode): Likewise.
18766 (eh_return_filter_mode): Return a scalar_int_mode rather than
18768 (libgcc_cmp_return_mode): Likewise.
18769 (libgcc_shift_count_mode): Likewise.
18770 (unwind_word_mode): Likewise.
18771 (addr_space.pointer_mode): Likewise.
18772 (addr_space.address_mode): Likewise.
18773 * doc/tm.texi: Regenerate.
18774 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
18776 (do_jump): Use scalar_int_mode for local variables.
18777 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
18778 rather than a machine_mode.
18779 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
18780 (scompare_loc_descriptor_wide): Likewise.
18781 (scompare_loc_descriptor_narrow): Likewise.
18782 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
18784 * except.c (sjlj_emit_dispatch_table): Likewise.
18785 (expand_builtin_eh_copy_values): Likewise.
18786 * explow.c (convert_memory_address_addr_space_1): Likewise.
18787 Take a scalar_int_mode rather than a machine_mode.
18788 (convert_memory_address_addr_space): Take a scalar_int_mode rather
18789 than a machine_mode.
18790 (memory_address_addr_space): Use scalar_int_mode for local variables.
18791 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
18792 rather than a machine_mode.
18793 * expmed.c (mask_rtx): Likewise.
18794 (init_expmed_one_conv): Likewise.
18795 (expand_mult_highpart_adjust): Likewise.
18796 (extract_high_half): Likewise.
18797 (expmed_mult_highpart_optab): Likewise.
18798 (expmed_mult_highpart): Likewise.
18799 (expand_smod_pow2): Likewise.
18800 (expand_sdiv_pow2): Likewise.
18801 (emit_store_flag_int): Likewise.
18802 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
18804 (extract_low_bits): Likewise.
18805 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18807 * expr.c (pieces_addr::adjust): Likewise.
18808 (can_store_by_pieces): Likewise.
18809 (store_by_pieces): Likewise.
18810 (clear_by_pieces_1): Likewise.
18811 (expand_expr_addr_expr_1): Likewise.
18812 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18813 (expand_expr_real_1): Likewise.
18814 (try_casesi): Likewise.
18815 * final.c (shorten_branches): Likewise.
18816 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18817 type of "mode" to machine_mode.
18818 * internal-fn.c (expand_arith_overflow_result_store): Take a
18819 scalar_int_mode rather than a machine_mode.
18820 (expand_mul_overflow): Use scalar_int_mode for local variables.
18821 * loop-doloop.c (doloop_modify): Likewise.
18822 (doloop_optimize): Likewise.
18823 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18824 than a machine_mode.
18825 (expand_doubleword_shift_condmove): Likewise.
18826 (expand_doubleword_shift): Likewise.
18827 (expand_doubleword_clz): Likewise.
18828 (expand_doubleword_popcount): Likewise.
18829 (expand_doubleword_parity): Likewise.
18830 (expand_absneg_bit): Use scalar_int_mode for local variables.
18831 (prepare_float_lib_cmp): Likewise.
18832 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18833 rather than a machine_mode.
18834 (convert_memory_address_addr_space): Likewise.
18835 (get_mode_bounds): Likewise.
18836 (get_address_mode): Return a scalar_int_mode rather than a
18838 * rtlanal.c (get_address_mode): Likewise.
18839 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18840 than a machine_mode.
18841 * targhooks.c (default_mode_rep_extended): Likewise.
18842 (default_valid_pointer_mode): Likewise.
18843 (default_addr_space_valid_pointer_mode): Likewise.
18844 (default_eh_return_filter_mode): Return a scalar_int_mode rather
18845 than a machine_mode.
18846 (default_libgcc_cmp_return_mode): Likewise.
18847 (default_libgcc_shift_count_mode): Likewise.
18848 (default_unwind_word_mode): Likewise.
18849 (default_addr_space_pointer_mode): Likewise.
18850 (default_addr_space_address_mode): Likewise.
18851 * targhooks.h (default_eh_return_filter_mode): Likewise.
18852 (default_libgcc_cmp_return_mode): Likewise.
18853 (default_libgcc_shift_count_mode): Likewise.
18854 (default_unwind_word_mode): Likewise.
18855 (default_addr_space_pointer_mode): Likewise.
18856 (default_addr_space_address_mode): Likewise.
18857 (default_mode_rep_extended): Take a scalar_int_mode rather than
18859 (default_valid_pointer_mode): Likewise.
18860 (default_addr_space_valid_pointer_mode): Likewise.
18861 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18863 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18864 rather than a machine_mode.
18865 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18866 for local variables.
18867 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18868 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18869 than a machine_mode.
18871 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18872 Alan Hayward <alan.hayward@arm.com>
18873 David Sherwood <david.sherwood@arm.com>
18875 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18876 the mode argument to scalar_int_mode.
18877 (do_jump_by_parts_zero_rtx): Likewise.
18878 (do_jump_by_parts_equality_rtx): Likewise.
18879 (do_jump_by_parts_greater): Take a mode argument.
18880 (do_jump_by_parts_equality): Likewise.
18881 (do_jump_1): Update calls accordingly.
18883 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18884 Alan Hayward <alan.hayward@arm.com>
18885 David Sherwood <david.sherwood@arm.com>
18887 * is-a.h (safe_dyn_cast): New function.
18888 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18889 (jump_table_for_label): Likewise.
18890 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18891 instead of an rtx_insn *.
18892 (shorten_branches): Use dyn_cast instead of LABEL_P and
18893 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18894 rtx_jump_table_data::get_data_mode.
18895 (final_scan_insn): Likewise.
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 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18902 trying to combine a full-register integer set with a subreg
18905 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18906 Alan Hayward <alan.hayward@arm.com>
18907 David Sherwood <david.sherwood@arm.com>
18909 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18910 that is always either address_mode or pointer_mode.
18912 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18913 Alan Hayward <alan.hayward@arm.com>
18914 David Sherwood <david.sherwood@arm.com>
18916 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
18917 when the two are known to be equal.
18919 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18920 Alan Hayward <alan.hayward@arm.com>
18921 David Sherwood <david.sherwood@arm.com>
18923 * simplify-rtx.c (simplify_const_unary_operation): Use
18924 is_a <scalar_int_mode> instead of checking for a nonzero
18925 precision. Forcibly convert op_mode to a scalar_int_mode
18926 in that case. More clearly differentiate the operand and
18927 result modes and use the former when deciding what the value
18928 of a count-bits operation should be. Use is_int_mode instead
18929 of checking for a MODE_INT. Remove redundant check for whether
18930 this mode has a zero precision.
18932 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18933 Alan Hayward <alan.hayward@arm.com>
18934 David Sherwood <david.sherwood@arm.com>
18936 * optabs.c (widen_leading): Change the type of the mode argument
18937 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
18938 (widen_bswap): Likewise.
18939 (expand_parity): Likewise.
18940 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
18941 (expand_ffs): Likewise.
18942 (epand_unop): Check for scalar integer modes before calling the
18945 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18946 Alan Hayward <alan.hayward@arm.com>
18947 David Sherwood <david.sherwood@arm.com>
18949 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
18951 (expand_expr_real_1): Update call accordingly.
18953 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18954 Alan Hayward <alan.hayward@arm.com>
18955 David Sherwood <david.sherwood@arm.com>
18957 * expmed.c (store_bit_field_using_insv): Add op0_mode and
18958 value_mode arguments. Use scalar_int_mode internally.
18959 (store_bit_field_1): Rename the new integer mode from imode
18960 to op0_mode and use it instead of GET_MODE (op0). Update calls
18961 to store_split_bit_field, store_bit_field_using_insv and
18962 store_fixed_bit_field.
18963 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
18964 Use scalar_int_mode internally. Use a bit count rather than a mode
18965 when calculating the largest bit size for get_best_mode.
18966 Update calls to store_split_bit_field and store_fixed_bit_field_1.
18967 (store_fixed_bit_field_1): Add mode and value_mode arguments.
18968 Remove assertion that OP0 has a scalar integer mode.
18969 (store_split_bit_field): Add op0_mode and value_mode arguments.
18970 Update calls to extract_fixed_bit_field.
18971 (extract_bit_field_using_extv): Add an op0_mode argument.
18972 Use scalar_int_mode internally.
18973 (extract_bit_field_1): Rename the new integer mode from imode to
18974 op0_mode and use it instead of GET_MODE (op0). Update calls to
18975 extract_split_bit_field, extract_bit_field_using_extv and
18976 extract_fixed_bit_field.
18977 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
18978 to extract_split_bit_field and extract_fixed_bit_field_1.
18979 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
18980 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
18981 on the target mode.
18982 (extract_split_bit_field): Add an op0_mode argument. Update call
18983 to extract_fixed_bit_field.
18985 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18986 Alan Hayward <alan.hayward@arm.com>
18987 David Sherwood <david.sherwood@arm.com>
18989 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
18990 * explow.c (hard_function_value): Likewise.
18991 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
18992 convert_to_mode call outside the loop.
18993 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
18994 for the mode iterator. Require the mode specified by max_pieces
18996 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
18998 (copy_blkmode_to_reg): Likewise.
18999 (set_storage_via_setmem): Likewise.
19000 * optabs.c (prepare_cmp_insn): Likewise.
19001 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19002 * stor-layout.c (finish_bitfield_representative): Likewise.
19004 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19005 Alan Hayward <alan.hayward@arm.com>
19006 David Sherwood <david.sherwood@arm.com>
19008 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
19009 * expr.c (convert_move): Use them.
19010 (convert_modes): Likewise.
19011 (store_expr_with_bounds): Likewise.
19013 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19014 Alan Hayward <alan.hayward@arm.com>
19015 David Sherwood <david.sherwood@arm.com>
19017 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
19018 parameter for the mode of "x". Remove the "known_x", "known_mode"
19019 and "known_ret" arguments. Change the type of the mode argument
19020 to scalar_int_mode.
19021 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
19022 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
19023 (reg_num_sign_bit_copies_for_combine): Likewise.
19024 * rtlanal.c (nonzero_bits1): Likewise.
19025 (num_sign_bit_copies1): Likewise.
19026 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
19027 (reg_num_sign_bit_copies_general): Likewise.
19028 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
19029 (reg_nonzero_bits_general): Likewise.
19031 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19032 Alan Hayward <alan.hayward@arm.com>
19033 David Sherwood <david.sherwood@arm.com>
19035 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
19036 than in subroutines. Return 1 for non-integer modes.
19037 (cached_num_sign_bit_copies): Change the type of the mode parameter
19038 to scalar_int_mode.
19039 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
19040 classes. Handle CONST_INT_P first and then check whether X also
19041 has a scalar integer mode. Check the same thing for inner registers
19042 of a SUBREG and for values that are being extended or truncated.
19044 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19045 Alan Hayward <alan.hayward@arm.com>
19046 David Sherwood <david.sherwood@arm.com>
19048 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
19049 in subroutines. Return the mode mask for non-integer modes.
19050 (cached_nonzero_bits): Change the type of the mode parameter
19051 to scalar_int_mode.
19052 (nonzero_bits1): Likewise. Remove early exit for other mode
19053 classes. Handle CONST_INT_P first and then check whether X
19054 also has a scalar integer mode.
19056 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19057 Alan Hayward <alan.hayward@arm.com>
19058 David Sherwood <david.sherwood@arm.com>
19060 * expr.c (widest_int_mode_for_size): Make the comment match the code.
19061 Return a scalar_int_mode and assert that the size is greater than
19063 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
19064 (op_by_pieces_d::op_by_pieces_d): Likewise.
19065 (op_by_pieces_d::run): Likewise.
19066 (can_store_by_pieces): Likewise.
19068 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19069 Alan Hayward <alan.hayward@arm.com>
19070 David Sherwood <david.sherwood@arm.com>
19072 * combine.c (extract_left_shift): Add a mode argument and update
19074 (make_compound_operation_int): Change the type of the mode parameter
19075 to scalar_int_mode and update the call to extract_left_shift.
19077 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19078 Alan Hayward <alan.hayward@arm.com>
19079 David Sherwood <david.sherwood@arm.com>
19081 * combine.c (simplify_and_const_int): Change the type of the mode
19082 parameter to scalar_int_mode.
19083 (simplify_and_const_int_1): Likewise. Update recursive call.
19085 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19086 Alan Hayward <alan.hayward@arm.com>
19087 David Sherwood <david.sherwood@arm.com>
19089 * combine.c (simplify_compare_const): Check that the mode is a
19090 scalar_int_mode (rather than VOIDmode) before testing its
19092 (simplify_comparison): Move COMPARISON_P handling out of the
19093 loop and restrict the latter part of the loop to scalar_int_modes.
19094 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
19095 and when considering SUBREG_REGs. Use is_int_mode instead of
19096 checking GET_MODE_CLASS against MODE_INT.
19098 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19099 Alan Hayward <alan.hayward@arm.com>
19100 David Sherwood <david.sherwood@arm.com>
19102 * combine.c (try_widen_shift_mode): Move check for equal modes to...
19103 (simplify_shift_const_1): ...here. Use scalar_int_mode for
19104 shift_unit_mode and for modes involved in scalar shifts.
19106 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19107 Alan Hayward <alan.hayward@arm.com>
19108 David Sherwood <david.sherwood@arm.com>
19110 * combine.c (force_int_to_mode): New function, split out from...
19111 (force_to_mode): ...here. Keep xmode up-to-date and use it
19112 instead of GET_MODE (x).
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 * optabs-query.h (extraction_insn::struct_mode): Change type to
19119 opt_scalar_int_mode and update comment.
19120 (extraction_insn::field_mode): Change type to scalar_int_mode.
19121 (extraction_insn::pos_mode): Likewise.
19122 * combine.c (make_extraction): Update accordingly.
19123 * optabs-query.c (get_traditional_extraction_insn): Likewise.
19124 (get_optab_extraction_insn): Likewise.
19125 * recog.c (simplify_while_replacing): Likewise.
19126 * expmed.c (narrow_bit_field_mem): Change the type of the mode
19127 parameter to opt_scalar_int_mode.
19129 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19130 Alan Hayward <alan.hayward@arm.com>
19131 David Sherwood <david.sherwood@arm.com>
19133 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
19134 to a scalar_int_mode instead of a machine_mode.
19135 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
19136 (get_best_mode): Return a boolean and use a pointer argument to store
19137 the selected mode. Replace the limit mode parameter with a bit limit.
19138 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
19139 for the values returned by bit_field_mode_iterator::next_mode.
19140 (store_bit_field): Update call to get_best_mode.
19141 (store_fixed_bit_field): Likewise.
19142 (extract_fixed_bit_field): Likewise.
19143 * expr.c (optimize_bitfield_assignment_op): Likewise.
19144 * fold-const.c (optimize_bit_field_compare): Likewise.
19145 (fold_truth_andor_1): Likewise.
19146 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
19147 Update for new type of m_mode.
19148 (get_best_mode): As above.
19150 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19151 Alan Hayward <alan.hayward@arm.com>
19152 David Sherwood <david.sherwood@arm.com>
19154 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
19155 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
19156 (store_bit_field): Check is_a <scalar_int_mode> before calling
19157 strict_volatile_bitfield_p.
19158 (extract_bit_field): Likewise.
19160 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19161 Alan Hayward <alan.hayward@arm.com>
19162 David Sherwood <david.sherwood@arm.com>
19164 * target.def (cstore_mode): Return a scalar_int_mode.
19165 * doc/tm.texi: Regenerate.
19166 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
19167 * targhooks.h (default_cstore_mode): Likewise.
19168 * targhooks.c (default_cstore_mode): Likewise, using a forced
19170 * expmed.c (emit_cstore): Expect the target of the cstore to be
19173 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19174 Alan Hayward <alan.hayward@arm.com>
19175 David Sherwood <david.sherwood@arm.com>
19177 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
19179 (niter_desc): Likewise mode.
19180 (iv_analyze): Add a mode parameter.
19182 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
19183 and change its type to scalar_int_mode.
19184 * loop-iv.c: Update commentary at head of file.
19185 (iv_constant): Pass the mode paraeter before the rtx it describes
19186 and change its type to scalar_int_mode. Remove VOIDmode handling.
19187 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
19188 (iv_extend): Likewise.
19189 (shorten_into_mode): Likewise.
19190 (iv_add): Use scalar_int_mode.
19191 (iv_mult): Likewise.
19192 (iv_shift): Likewise.
19193 (canonicalize_iv_subregs): Likewise.
19194 (get_biv_step_1): Pass the outer_mode parameter before the rtx
19195 it describes and change its mode to scalar_int_mode. Also change
19196 the type of the returned inner_mode to scalar_int_mode.
19197 (get_biv_step): Likewise, turning outer_mode from a pointer
19198 into a direct parameter. Update call to get_biv_step_1.
19199 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
19200 iv_constant and get_biv_step.
19201 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
19202 and change its type to scalar_int_mode. Don't initialise iv->mode
19203 to VOIDmode and remove later checks for its still being VOIDmode.
19204 Update calls to iv_analyze_op and iv_analyze_expr. Check
19205 is_a <scalar_int_mode> when changing the mode under consideration.
19206 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
19207 Update call to iv_analyze_expr.
19208 (iv_analyze_op): Add a mode parameter. Reject subregs whose
19209 inner register is not also a scalar_int_mode. Update call to
19211 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
19212 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
19213 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
19214 separate mode class checks. Update calls to iv_analyze. Remove
19215 fix-up of VOIDmodes after iv_analyze_biv.
19216 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
19217 don't have a scalar_int_mode. Update call to biv_p.
19219 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19220 Alan Hayward <alan.hayward@arm.com>
19221 David Sherwood <david.sherwood@arm.com>
19223 * cfgexpand.c (convert_debug_memory_address): Use
19224 as_a <scalar_int_mode>.
19225 * combine.c (expand_compound_operation): Likewise.
19226 (make_extraction): Likewise.
19227 (change_zero_ext): Likewise.
19228 (simplify_comparison): Likewise.
19229 * cse.c (cse_insn): Likewise.
19230 * dwarf2out.c (minmax_loc_descriptor): Likewise.
19231 (mem_loc_descriptor): Likewise.
19232 (loc_descriptor): Likewise.
19233 * expmed.c (init_expmed_one_mode): Likewise.
19234 (synth_mult): Likewise.
19235 (emit_store_flag_1): Likewise.
19236 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
19237 of a comparison with size.
19238 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
19239 (reduce_to_bit_field_precision): Likewise.
19240 * function.c (expand_function_end): Likewise.
19241 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
19242 * loop-doloop.c (doloop_modify): Likewise.
19243 * optabs.c (expand_binop): Likewise.
19244 (expand_unop): Likewise.
19245 (expand_copysign_absneg): Likewise.
19246 (prepare_cmp_insn): Likewise.
19247 (maybe_legitimize_operand): Likewise.
19248 * recog.c (const_scalar_int_operand): Likewise.
19249 * rtlanal.c (get_address_mode): Likewise.
19250 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19251 (simplify_cond_clz_ctz): Likewise.
19252 * tree-nested.c (get_nl_goto_field): Likewise.
19253 * tree.c (build_vector_type_for_mode): Likewise.
19254 * var-tracking.c (use_narrower_mode): Likewise.
19256 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19257 Alan Hayward <alan.hayward@arm.com>
19258 David Sherwood <david.sherwood@arm.com>
19260 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
19261 * builtins.c (expand_builtin_signbit): Use it.
19262 * cfgexpand.c (expand_debug_expr): Likewise.
19263 * dojump.c (do_jump): Likewise.
19264 (do_compare_and_jump): Likewise.
19265 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
19266 * expmed.c (make_tree): Likewise.
19267 * expr.c (expand_expr_real_2): Likewise.
19268 (expand_expr_real_1): Likewise.
19269 (try_casesi): Likewise.
19270 * fold-const-call.c (fold_const_call_ss): Likewise.
19271 * fold-const.c (unextend): Likewise.
19272 (extract_muldiv_1): Likewise.
19273 (fold_single_bit_test): Likewise.
19274 (native_encode_int): Likewise.
19275 (native_encode_string): Likewise.
19276 (native_interpret_int): Likewise.
19277 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
19278 * internal-fn.c (expand_addsub_overflow): Likewise.
19279 (expand_neg_overflow): Likewise.
19280 (expand_mul_overflow): Likewise.
19281 (expand_arith_overflow): Likewise.
19282 * match.pd: Likewise.
19283 * stor-layout.c (layout_type): Likewise.
19284 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
19285 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
19286 * tree-ssanames.c (get_range_info): Likewise.
19287 * tree-switch-conversion.c (array_value_type) Likewise.
19288 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
19289 (vect_recog_divmod_pattern): Likewise.
19290 (vect_recog_mixed_size_cond_pattern): Likewise.
19291 * tree-vrp.c (extract_range_basic): Likewise.
19292 (simplify_float_conversion_using_ranges): Likewise.
19293 * tree.c (int_fits_type_p): Likewise.
19294 * ubsan.c (instrument_bool_enum_load): Likewise.
19295 * varasm.c (mergeable_string_section): Likewise.
19296 (narrowing_initializer_constant_valid_p): Likewise.
19297 (output_constant): Likewise.
19299 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19300 Alan Hayward <alan.hayward@arm.com>
19301 David Sherwood <david.sherwood@arm.com>
19303 * machmode.h (NARROWEST_INT_MODE): New macro.
19304 * expr.c (alignment_for_piecewise_move): Use it instead of
19305 GET_CLASS_NARROWEST_MODE (MODE_INT).
19306 (push_block): Likewise.
19307 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
19309 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19311 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19312 Alan Hayward <alan.hayward@arm.com>
19313 David Sherwood <david.sherwood@arm.com>
19315 * postreload.c (move2add_valid_value_p): Change the type of the
19316 mode parameter to scalar_int_mode.
19317 (move2add_use_add2_insn): Add a mode parameter and use it instead
19319 (move2add_use_add3_insn): Likewise.
19320 (reload_cse_move2add): Update accordingly.
19322 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19323 Alan Hayward <alan.hayward@arm.com>
19324 David Sherwood <david.sherwood@arm.com>
19326 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
19328 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
19329 * optabs.c (expand_unop): Likewise.
19331 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19332 Alan Hayward <alan.hayward@arm.com>
19333 David Sherwood <david.sherwood@arm.com>
19335 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
19336 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
19337 (popcount_loc_descriptor): Likewise.
19338 (bswap_loc_descriptor): Likewise.
19339 (rotate_loc_descriptor): Likewise.
19340 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
19341 calling the functions above.
19343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19344 Alan Hayward <alan.hayward@arm.com>
19345 David Sherwood <david.sherwood@arm.com>
19347 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
19349 (try_combine): Likewise.
19350 (simplify_if_then_else): Likewise.
19351 * cse.c (cse_insn): Likewise.
19352 * dwarf2out.c (mem_loc_descriptor): Likewise.
19353 * emit-rtl.c (gen_lowpart_common): Likewise.
19354 * simplify-rtx.c (simplify_truncation): Likewise.
19355 (simplify_binary_operation_1): Likewise.
19356 (simplify_const_relational_operation): Likewise.
19357 (simplify_ternary_operation): Likewise.
19358 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
19360 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19361 Alan Hayward <alan.hayward@arm.com>
19362 David Sherwood <david.sherwood@arm.com>
19364 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
19365 * reload.c (push_reload): Likewise.
19366 (find_reloads): Likewise.
19368 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19369 Alan Hayward <alan.hayward@arm.com>
19370 David Sherwood <david.sherwood@arm.com>
19372 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
19373 (make_compound_operation_int): Likewise.
19374 (change_zero_ext): Likewise.
19375 * expr.c (convert_move): Likewise.
19376 (convert_modes): Likewise.
19377 * fwprop.c (forward_propagate_subreg): Likewise.
19378 * loop-iv.c (get_biv_step_1): Likewise.
19379 * optabs.c (widen_operand): Likewise.
19380 * postreload.c (move2add_valid_value_p): Likewise.
19381 * recog.c (simplify_while_replacing): Likewise.
19382 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19383 (simplify_binary_operation_1): Likewise. Remove redundant
19384 mode equality check.
19386 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19387 Alan Hayward <alan.hayward@arm.com>
19388 David Sherwood <david.sherwood@arm.com>
19390 * combine.c (combine_simplify_rtx): Add checks for
19391 is_a <scalar_int_mode>.
19392 (simplify_if_then_else): Likewise.
19393 (make_field_assignment): Likewise.
19394 (simplify_comparison): Likewise.
19395 * ifcvt.c (noce_try_bitop): Likewise.
19396 * loop-invariant.c (canonicalize_address_mult): Likewise.
19397 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19399 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19400 Alan Hayward <alan.hayward@arm.com>
19401 David Sherwood <david.sherwood@arm.com>
19403 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
19404 is_a <scalar_int_mode> instead of != BLKmode.
19406 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19407 Alan Hayward <alan.hayward@arm.com>
19408 David Sherwood <david.sherwood@arm.com>
19410 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
19411 instead of != VOIDmode.
19412 * combine.c (if_then_else_cond): Likewise.
19413 (change_zero_ext): Likewise.
19414 * dwarf2out.c (mem_loc_descriptor): Likewise.
19415 (loc_descriptor): Likewise.
19416 * rtlanal.c (canonicalize_condition): Likewise.
19417 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
19419 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19420 Alan Hayward <alan.hayward@arm.com>
19421 David Sherwood <david.sherwood@arm.com>
19423 * simplify-rtx.c (simplify_binary_operation_1): Use
19424 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
19426 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19427 Alan Hayward <alan.hayward@arm.com>
19428 David Sherwood <david.sherwood@arm.com>
19430 * wide-int.h (int_traits<unsigned char>) New class.
19431 (int_traits<unsigned short>) Likewise.
19432 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
19433 Use GET_MODE_UNIT_PRECISION and remove redundant test for
19435 * combine.c (set_nonzero_bits_and_sign_copies): Use
19436 is_a <scalar_int_mode>.
19437 (find_split_point): Likewise.
19438 (combine_simplify_rtx): Likewise.
19439 (simplify_logical): Likewise.
19440 (expand_compound_operation): Likewise.
19441 (expand_field_assignment): Likewise.
19442 (make_compound_operation): Likewise.
19443 (extended_count): Likewise.
19444 (change_zero_ext): Likewise.
19445 (simplify_comparison): Likewise.
19446 * dwarf2out.c (scompare_loc_descriptor): Likewise.
19447 (ucompare_loc_descriptor): Likewise.
19448 (minmax_loc_descriptor): Likewise.
19449 (mem_loc_descriptor): Likewise.
19450 (loc_descriptor): Likewise.
19451 * expmed.c (init_expmed_one_mode): Likewise.
19452 * lra-constraints.c (lra_constraint_offset): Likewise.
19453 * optabs.c (prepare_libcall_arg): Likewise.
19454 * postreload.c (move2add_note_store): Likewise.
19455 * reload.c (operands_match_p): Likewise.
19456 * rtl.h (load_extend_op): Likewise.
19457 * rtlhooks.c (gen_lowpart_general): Likewise.
19458 * simplify-rtx.c (simplify_truncation): Likewise.
19459 (simplify_unary_operation_1): Likewise.
19460 (simplify_binary_operation_1): Likewise.
19461 (simplify_const_binary_operation): Likewise.
19462 (simplify_const_relational_operation): Likewise.
19463 (simplify_subreg): Likewise.
19464 * stor-layout.c (bitwise_mode_for_mode): Likewise.
19465 * var-tracking.c (adjust_mems): Likewise.
19466 (prepare_call_arguments): Likewise.
19468 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19469 Alan Hayward <alan.hayward@arm.com>
19470 David Sherwood <david.sherwood@arm.com>
19472 * machmode.h (is_int_mode): New fuction.
19473 * combine.c (find_split_point): Use it.
19474 (combine_simplify_rtx): Likewise.
19475 (simplify_if_then_else): Likewise.
19476 (simplify_set): Likewise.
19477 (simplify_shift_const_1): Likewise.
19478 (simplify_comparison): Likewise.
19479 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
19480 * cse.c (notreg_cost): Likewise.
19481 (cse_insn): Likewise.
19482 * cselib.c (cselib_lookup_1): Likewise.
19483 * dojump.c (do_jump_1): Likewise.
19484 (do_compare_rtx_and_jump): Likewise.
19485 * dse.c (get_call_args): Likewise.
19486 * dwarf2out.c (rtl_for_decl_init): Likewise.
19487 (native_encode_initializer): Likewise.
19488 * expmed.c (emit_store_flag_1): Likewise.
19489 (emit_store_flag): Likewise.
19490 * expr.c (convert_modes): Likewise.
19491 (store_field): Likewise.
19492 (expand_expr_real_1): Likewise.
19493 * fold-const.c (fold_read_from_constant_string): Likewise.
19494 * gimple-ssa-sprintf.c (get_format_string): Likewise.
19495 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
19496 * optabs.c (expand_binop): Likewise.
19497 (expand_unop): Likewise.
19498 (expand_abs_nojump): Likewise.
19499 (expand_one_cmpl_abs_nojump): Likewise.
19500 * simplify-rtx.c (mode_signbit_p): Likewise.
19501 (val_signbit_p): Likewise.
19502 (val_signbit_known_set_p): Likewise.
19503 (val_signbit_known_clear_p): Likewise.
19504 (simplify_relational_operation_1): Likewise.
19505 * tree.c (vector_type_mode): Likewise.
19507 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19508 Alan Hayward <alan.hayward@arm.com>
19509 David Sherwood <david.sherwood@arm.com>
19511 * machmode.h (smallest_mode_for_size): Fix formatting.
19512 (smallest_int_mode_for_size): New function.
19513 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
19514 instead of smallest_mode_for_size.
19515 * combine.c (make_extraction): Likewise.
19516 * config/arc/arc.c (arc_expand_movmem): Likewise.
19517 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
19518 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
19519 * config/s390/s390.c (s390_expand_insv): Likewise.
19520 * config/sparc/sparc.c (assign_int_registers): Likewise.
19521 * config/spu/spu.c (spu_function_value): Likewise.
19522 (spu_function_arg): Likewise.
19523 * coverage.c (get_gcov_type): Likewise.
19524 (get_gcov_unsigned_t): Likewise.
19525 * dse.c (find_shift_sequence): Likewise.
19526 * expmed.c (store_bit_field_1): Likewise.
19527 * expr.c (convert_move): Likewise.
19528 (store_field): Likewise.
19529 * internal-fn.c (expand_arith_overflow): Likewise.
19530 * optabs-query.c (get_best_extraction_insn): Likewise.
19531 * optabs.c (expand_twoval_binop_libfunc): Likewise.
19532 * stor-layout.c (layout_type): Likewise.
19533 (initialize_sizetypes): Likewise.
19534 * targhooks.c (default_get_mask_mode): Likewise.
19535 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
19537 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19538 Alan Hayward <alan.hayward@arm.com>
19539 David Sherwood <david.sherwood@arm.com>
19541 * machmode.h (opt_mode::else_blk): New function.
19542 (int_mode_for_mode): Declare.
19543 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
19544 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
19546 * cfgexpand.c (expand_debug_expr): Likewise.
19547 * combine.c (gen_lowpart_or_truncate): Likewise.
19548 (gen_lowpart_for_combine): Likewise.
19549 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
19550 * config/avr/avr.c (avr_to_int_mode): Likewise.
19551 (avr_out_plus_1): Likewise.
19552 (avr_out_plus): Likewise.
19553 (avr_out_round): Likewise.
19554 * config/i386/i386.c (ix86_split_to_parts): Likewise.
19555 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
19556 (s390_expand_vcond): Likewise.
19557 * config/spu/spu.c (spu_split_immediate): Likewise.
19558 (spu_expand_mov): Likewise.
19559 * dse.c (get_stored_val): Likewise.
19560 * expmed.c (store_bit_field_1): Likewise.
19561 (convert_extracted_bit_field): Use int_mode_for_mode instead of
19563 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
19564 (extract_low_bits): Likewise.
19565 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
19566 handling rather than repeating the check.
19567 (emit_group_store): Likewise.
19568 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
19569 * optabs.c (expand_absneg_bit): Likewise.
19570 (expand_copysign_absneg): Likewise.
19571 (expand_copysign_bit): Likewise.
19572 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
19573 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
19574 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
19575 * var-tracking.c (prepare_call_arguments): Likewise.
19576 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
19577 int_mode_for_mode instead of mode_for_size.
19578 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
19580 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19581 Alan Hayward <alan.hayward@arm.com>
19582 David Sherwood <david.sherwood@arm.com>
19584 * machmode.h (int_mode_for_size): New function.
19585 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
19586 instead of mode_for_size.
19587 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
19589 * combine.c (expand_field_assignment): Use int_mode_for_size
19590 instead of mode_for_size.
19591 (make_extraction): Likewise.
19592 (simplify_shift_const_1): Likewise.
19593 (simplify_comparison): Likewise.
19594 * dojump.c (do_jump): Likewise.
19595 * dwarf2out.c (mem_loc_descriptor): Likewise.
19596 * emit-rtl.c (init_derived_machine_modes): Likewise.
19597 * expmed.c (flip_storage_order): Likewise.
19598 (convert_extracted_bit_field): Likewise.
19599 * expr.c (copy_blkmode_from_reg): Likewise.
19600 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
19601 * internal-fn.c (expand_mul_overflow): Likewise.
19602 * lower-subreg.c (simple_move): Likewise.
19603 * optabs-libfuncs.c (init_optabs): Likewise.
19604 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19605 * tree.c (vector_type_mode): Likewise.
19606 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
19607 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
19608 * tree-vect-generic.c (expand_vector_parallel): Likewise.
19609 * tree-vect-stmts.c (vectorizable_load): Likewise.
19610 (vectorizable_store): Likewise.
19612 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19613 Alan Hayward <alan.hayward@arm.com>
19614 David Sherwood <david.sherwood@arm.com>
19616 * coretypes.h (pod_mode): New type.
19617 (scalar_int_mode_pod): New typedef.
19618 * machmode.h (pod_mode): New class.
19619 (int_n_data_t::m): Change type to scalar_int_mode_pod.
19620 * genmodes.c (emit_mode_int_n): Update accordingly.
19621 * lower-subreg.h (target_lower_subreg): Change type to
19622 scalar_int_mode_pod.
19623 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
19624 scalar_int_mode_pod.
19626 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19627 Alan Hayward <alan.hayward@arm.com>
19628 David Sherwood <david.sherwood@arm.com>
19630 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
19631 machine_mode to scalar_int_mode.
19632 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
19633 (rs6000_option_override_internal): Remove cast to int.
19634 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
19635 machine_mode to scalar_int_mode.
19636 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
19637 (rs6000_option_override_internal): Remove cast to int.
19638 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
19639 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
19641 * config/s390/s390.c (s390_expand_builtin): Likewise.
19642 * coretypes.h (scalar_int_mode): New type.
19643 (opt_scalar_int_mode): New typedef.
19644 * machmode.h (scalar_int_mode): New class.
19645 (scalar_int_mode::includes_p): New function.
19646 (byte_mode): Change type to scalar_int_mode.
19647 (word_mode): Likewise.
19648 (ptr_mode): Likewise.
19649 * emit-rtl.c (byte_mode): Likewise.
19650 (word_mode): Likewise.
19651 (ptr_mode): Likewise.
19652 (init_derived_machine_modes): Update accordingly.
19653 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
19654 and MODE_PARTIAL_INT.
19655 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
19656 opt_scalar_int_mode.
19658 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19659 Alan Hayward <alan.hayward@arm.com>
19660 David Sherwood <david.sherwood@arm.com>
19662 * target.def (libgcc_floating_mode_supported_p): Take a
19664 * doc/tm.texi: Regenerate.
19665 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
19667 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19668 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
19671 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19672 Alan Hayward <alan.hayward@arm.com>
19673 David Sherwood <david.sherwood@arm.com>
19675 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
19676 * doc/tm.texi: Regenerate.
19677 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
19678 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
19679 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
19680 * targhooks.h (default_floatn_mode): Likewise.
19681 * targhooks.c (default_floatn_mode): Likewise.
19682 * tree.c (build_common_tree_nodes): Update accordingly.
19684 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19685 Alan Hayward <alan.hayward@arm.com>
19686 David Sherwood <david.sherwood@arm.com>
19688 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
19689 (mode_iterator::iterate_p): Likewise.
19690 (mode_iterator::get_wider): Likewise.
19691 * expr.c (init_expr_target): Use opt_scalar_float_mode.
19693 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19694 Alan Hayward <alan.hayward@arm.com>
19695 David Sherwood <david.sherwood@arm.com>
19697 * coretypes.h (opt_scalar_float_mode): New typedef.
19698 * machmode.h (float_mode_for_size): New function.
19699 * emit-rtl.c (double_mode): Delete.
19700 (init_emit_once): Use float_mode_for_size.
19701 * stor-layout.c (layout_type): Likewise.
19702 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
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 * output.h (assemble_real): Take a scalar_float_mode.
19709 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
19710 * config/arm/arm.md (consttable_4): Likewise.
19711 (consttable_8): Likewise.
19712 (consttable_16): Likewise.
19713 * config/mips/mips.md (consttable_float): Likewise.
19714 * config/s390/s390.c (s390_output_pool_entry): Likewise.
19715 * varasm.c (assemble_real): Take a scalar_float_mode.
19716 (output_constant_pool_2): Update accordingly.
19717 (output_constant): Likewise.
19719 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19720 Alan Hayward <alan.hayward@arm.com>
19721 David Sherwood <david.sherwood@arm.com>
19723 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
19724 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
19725 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
19726 (native_encode_real): Likewise.
19727 (native_interpret_real): Likewise.
19728 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
19729 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19731 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19732 Alan Hayward <alan.hayward@arm.com>
19733 David Sherwood <david.sherwood@arm.com>
19735 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
19736 <scalar_float_mode>. Simplify.
19737 (gen_extend_conv_libfunc): Likewise.
19739 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19740 Alan Hayward <alan.hayward@arm.com>
19741 David Sherwood <david.sherwood@arm.com>
19743 * coretypes.h (scalar_float_mode): New type.
19744 * machmode.h (mode_traits::from_int): Use machine_mode if
19745 USE_ENUM_MODES is defined.
19746 (is_a): New function.
19748 (dyn_cast): Likewise.
19749 (scalar_float_mode): New class.
19750 (scalar_float_mode::includes_p): New function.
19751 (is_float_mode): Likewise.
19752 * gdbhooks.py (MachineModePrinter): New class.
19753 (build_pretty_printer): Use it for scalar_float_mode.
19754 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
19755 (format_helper::format_helper): Turn into a template.
19756 * genmodes.c (get_mode_class): New function.
19757 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
19758 or machine_mode if none.
19759 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
19760 as_a <scalar_float_mode>.
19761 * dwarf2out.c (mem_loc_descriptor): Likewise.
19762 (insert_float): Likewise.
19763 (add_const_value_attribute): Likewise.
19764 * simplify-rtx.c (simplify_immed_subreg): Likewise.
19765 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
19766 (expand_unop): Update accordingly.
19767 (expand_abs_nojump): Likewise.
19768 (expand_copysign_absneg): Take a scalar_float_mode.
19769 (expand_copysign_bit): Likewise.
19770 (expand_copysign): Update accordingly.
19772 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19773 Alan Hayward <alan.hayward@arm.com>
19774 David Sherwood <david.sherwood@arm.com>
19776 * coretypes.h (opt_mode): New class.
19777 * machmode.h (opt_mode): Likewise.
19778 (opt_mode::else_void): New function.
19779 (opt_mode::require): Likewise.
19780 (opt_mode::exists): Likewise.
19781 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
19782 (GET_MODE_2XWIDER_MODE): Likewise.
19783 (mode_iterator::get_wider): Update accordingly.
19784 (mode_iterator::get_2xwider): Likewise.
19785 (mode_iterator::get_known_wider): Likewise, turning into a template.
19786 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
19787 forcing a wider mode to exist.
19788 * config/cr16/cr16.h (LONG_REG_P): Likewise.
19789 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19790 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
19791 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
19792 * lower-subreg.c (init_lower_subreg): Likewise.
19793 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
19794 on the final iteration.
19795 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
19796 a wider mode exists before asking for a move pattern.
19797 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
19798 forcing a wider mode to exist.
19799 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
19800 returning false if no such mode exists.
19801 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
19802 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
19803 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
19804 Avoid checking for a MODE_INT if we already know the mode is not a
19806 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19807 forcing a wider mode to exist.
19808 (expmed_mult_highpart_optab): Likewise.
19809 (expmed_mult_highpart): Likewise.
19810 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19812 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19813 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19814 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19815 * internal-fn.c (expand_mul_overflow): Update use of
19816 GET_MODE_2XWIDER_MODE.
19817 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19818 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19819 GET_MODE_WIDER_MODE.
19820 (convert_plusminus_to_widen): Likewise.
19821 * tree-switch-conversion.c (array_value_type): Likewise.
19822 * var-tracking.c (emit_note_insn_var_location): Likewise.
19823 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19824 Return false inside rather than outside the loop if no wider mode
19826 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19827 and GET_MODE_2XWIDER_MODE
19828 (can_compare_p): Use else_void.
19829 * gdbhooks.py (OptMachineModePrinter): New class.
19830 (build_pretty_printer): Use it for opt_mode.
19832 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19833 Alan Hayward <alan.hayward@arm.com>
19834 David Sherwood <david.sherwood@arm.com>
19836 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19837 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19839 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19840 Alan Hayward <alan.hayward@arm.com>
19841 David Sherwood <david.sherwood@arm.com>
19843 * machmode.h (mode_traits): New structure.
19844 (get_narrowest_mode): New function.
19845 (mode_iterator::start): Likewise.
19846 (mode_iterator::iterate_p): Likewise.
19847 (mode_iterator::get_wider): Likewise.
19848 (mode_iterator::get_known_wider): Likewise.
19849 (mode_iterator::get_2xwider): Likewise.
19850 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19851 (FOR_EACH_MODE): Likewise.
19852 (FOR_EACH_MODE_FROM): Likewise.
19853 (FOR_EACH_MODE_UNTIL): Likewise.
19854 (FOR_EACH_WIDER_MODE): Likewise.
19855 (FOR_EACH_2XWIDER_MODE): Likewise.
19856 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19857 * combine.c (simplify_comparison): Likewise
19858 * config/i386/i386.c (type_natural_mode): Likewise.
19859 * cse.c (cse_insn): Likewise.
19860 * dse.c (find_shift_sequence): Likewise.
19861 * emit-rtl.c (init_derived_machine_modes): Likewise.
19862 (init_emit_once): Likewise.
19863 * explow.c (hard_function_value): Likewise.
19864 * expmed.c (extract_fixed_bit_field_1): Likewise.
19865 (extract_bit_field_1): Likewise.
19866 (expand_divmod): Likewise.
19867 (emit_store_flag_1): Likewise.
19868 * expr.c (init_expr_target): Likewise.
19869 (convert_move): Likewise.
19870 (alignment_for_piecewise_move): Likewise.
19871 (widest_int_mode_for_size): Likewise.
19872 (emit_block_move_via_movmem): Likewise.
19873 (copy_blkmode_to_reg): Likewise.
19874 (set_storage_via_setmem): Likewise.
19875 (compress_float_constant): Likewise.
19876 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19877 * optabs-query.c (get_best_extraction_insn): Likewise.
19878 * optabs.c (expand_binop): Likewise.
19879 (expand_twoval_unop): Likewise.
19880 (expand_twoval_binop): Likewise.
19881 (widen_leading): Likewise.
19882 (widen_bswap): Likewise.
19883 (expand_parity): Likewise.
19884 (expand_unop): Likewise.
19885 (prepare_cmp_insn): Likewise.
19886 (prepare_float_lib_cmp): Likewise.
19887 (expand_float): Likewise.
19888 (expand_fix): Likewise.
19889 (expand_sfix_optab): Likewise.
19890 * postreload.c (move2add_use_add2_insn): Likewise.
19891 * reg-stack.c (reg_to_stack): Likewise.
19892 * reginfo.c (choose_hard_reg_mode): Likewise.
19893 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19894 * stor-layout.c (mode_for_size): Likewise.
19895 (smallest_mode_for_size): Likewise.
19896 (mode_for_vector): Likewise.
19897 (finish_bitfield_representative): Likewise.
19898 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19899 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19900 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19901 * var-tracking.c (prepare_call_arguments): Likewise.
19903 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19904 Alan Hayward <alan.hayward@arm.com>
19905 David Sherwood <david.sherwood@arm.com>
19907 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19908 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19909 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19910 * machmode.h (mode_size): Move earlier in file.
19911 (mode_precision): Likewise.
19912 (mode_inner): Likewise.
19913 (mode_nunits): Likewise.
19914 (mode_unit_size): Likewise.
19915 (unit_unit_precision): Likewise.
19916 (mode_wider): Likewise.
19917 (mode_2xwider): Likewise.
19918 (machine_mode): New class.
19919 (mode_to_bytes): New function.
19920 (mode_to_bits): Likewise.
19921 (mode_to_precision): Likewise.
19922 (mode_to_inner): Likewise.
19923 (mode_to_unit_size): Likewise.
19924 (mode_to_unit_precision): Likewise.
19925 (mode_to_nunits): Likewise.
19926 (GET_MODE_SIZE): Use mode_to_bytes.
19927 (GET_MODE_BITSIZE): Use mode_to_bits.
19928 (GET_MODE_PRECISION): Use mode_to_precision.
19929 (GET_MODE_INNER): Use mode_to_inner.
19930 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
19931 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
19932 (GET_MODE_NUNITS): Use mode_to_nunits.
19933 * system.h (ALWAYS_INLINE): New macro.
19934 * config/powerpcspe/powerpcspe-c.c
19935 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
19936 int for arg1_mode and arg2_mode.
19938 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19939 Alan Hayward <alan.hayward@arm.com>
19940 David Sherwood <david.sherwood@arm.com>
19942 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19943 Prefix mode names with E_ in case statements.
19944 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19945 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
19946 (aarch64_split_simd_move): Likewise.
19947 (aarch64_gen_storewb_pair): Likewise.
19948 (aarch64_gen_loadwb_pair): Likewise.
19949 (aarch64_gen_store_pair): Likewise.
19950 (aarch64_gen_load_pair): Likewise.
19951 (aarch64_get_condition_code_1): Likewise.
19952 (aarch64_constant_pool_reload_icode): Likewise.
19953 (get_rsqrte_type): Likewise.
19954 (get_rsqrts_type): Likewise.
19955 (get_recpe_type): Likewise.
19956 (get_recps_type): Likewise.
19957 (aarch64_gimplify_va_arg_expr): Likewise.
19958 (aarch64_simd_container_mode): Likewise.
19959 (aarch64_emit_load_exclusive): Likewise.
19960 (aarch64_emit_store_exclusive): Likewise.
19961 (aarch64_expand_compare_and_swap): Likewise.
19962 (aarch64_gen_atomic_cas): Likewise.
19963 (aarch64_emit_bic): Likewise.
19964 (aarch64_emit_atomic_swap): Likewise.
19965 (aarch64_emit_atomic_load_op): Likewise.
19966 (aarch64_evpc_trn): Likewise.
19967 (aarch64_evpc_uzp): Likewise.
19968 (aarch64_evpc_zip): Likewise.
19969 (aarch64_evpc_ext): Likewise.
19970 (aarch64_evpc_rev): Likewise.
19971 (aarch64_evpc_dup): Likewise.
19972 (aarch64_gen_ccmp_first): Likewise.
19973 (aarch64_gen_ccmp_next): Likewise.
19974 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
19975 (alpha_emit_xfloating_libcall): Likewise.
19976 (emit_insxl): Likewise.
19977 (alpha_arg_type): Likewise.
19978 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
19979 (arc_preferred_simd_mode): Likewise.
19980 (arc_secondary_reload): Likewise.
19981 (get_arc_condition_code): Likewise.
19982 (arc_print_operand): Likewise.
19983 (arc_legitimate_constant_p): Likewise.
19984 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19985 * config/arc/arc.md (casesi_load): Likewise.
19986 (casesi_compact_jump): Likewise.
19987 * config/arc/predicates.md (proper_comparison_operator): Likewise.
19988 (cc_use_register): Likewise.
19989 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19990 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
19991 (arm_init_iwmmxt_builtins): Likewise.
19992 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
19993 (neon_expand_vector_init): Likewise.
19994 (arm_attr_length_move_neon): Likewise.
19995 (maybe_get_arm_condition_code): Likewise.
19996 (arm_emit_vector_const): Likewise.
19997 (arm_preferred_simd_mode): Likewise.
19998 (arm_output_iwmmxt_tinsr): Likewise.
19999 (thumb1_output_casesi): Likewise.
20000 (thumb2_output_casesi): Likewise.
20001 (arm_emit_load_exclusive): Likewise.
20002 (arm_emit_store_exclusive): Likewise.
20003 (arm_expand_compare_and_swap): Likewise.
20004 (arm_evpc_neon_vuzp): Likewise.
20005 (arm_evpc_neon_vzip): Likewise.
20006 (arm_evpc_neon_vrev): Likewise.
20007 (arm_evpc_neon_vtrn): Likewise.
20008 (arm_evpc_neon_vext): Likewise.
20009 (arm_validize_comparison): Likewise.
20010 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
20011 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
20012 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
20013 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
20014 (c6x_preferred_simd_mode): Likewise.
20015 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
20016 (epiphany_rtx_costs): Likewise.
20017 * config/epiphany/predicates.md (proper_comparison_operator):
20019 * config/frv/frv.c (condexec_memory_operand): Likewise.
20020 (frv_emit_move): Likewise.
20021 (output_move_single): Likewise.
20022 (output_condmove_single): Likewise.
20023 (frv_hard_regno_mode_ok): Likewise.
20024 (frv_matching_accg_mode): Likewise.
20025 * config/h8300/h8300.c (split_adds_subs): Likewise.
20026 (h8300_rtx_costs): Likewise.
20027 (h8300_print_operand): Likewise.
20028 (compute_mov_length): Likewise.
20029 (output_logical_op): Likewise.
20030 (compute_logical_op_length): Likewise.
20031 (compute_logical_op_cc): Likewise.
20032 (h8300_shift_needs_scratch_p): Likewise.
20033 (output_a_shift): Likewise.
20034 (compute_a_shift_length): Likewise.
20035 (compute_a_shift_cc): Likewise.
20036 (expand_a_rotate): Likewise.
20037 (output_a_rotate): Likewise.
20038 * config/i386/i386.c (classify_argument): Likewise.
20039 (function_arg_advance_32): Likewise.
20040 (function_arg_32): Likewise.
20041 (function_arg_64): Likewise.
20042 (function_value_64): Likewise.
20043 (ix86_gimplify_va_arg): Likewise.
20044 (ix86_legitimate_constant_p): Likewise.
20045 (put_condition_code): Likewise.
20046 (split_double_mode): Likewise.
20047 (ix86_avx256_split_vector_move_misalign): Likewise.
20048 (ix86_expand_vector_logical_operator): Likewise.
20049 (ix86_split_idivmod): Likewise.
20050 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
20051 (ix86_build_const_vector): Likewise.
20052 (ix86_build_signbit_mask): Likewise.
20053 (ix86_match_ccmode): Likewise.
20054 (ix86_cc_modes_compatible): Likewise.
20055 (ix86_expand_branch): Likewise.
20056 (ix86_expand_sse_cmp): Likewise.
20057 (ix86_expand_sse_movcc): Likewise.
20058 (ix86_expand_int_sse_cmp): Likewise.
20059 (ix86_expand_vec_perm_vpermi2): Likewise.
20060 (ix86_expand_vec_perm): Likewise.
20061 (ix86_expand_sse_unpack): Likewise.
20062 (ix86_expand_int_addcc): Likewise.
20063 (ix86_split_to_parts): Likewise.
20064 (ix86_vectorize_builtin_gather): Likewise.
20065 (ix86_vectorize_builtin_scatter): Likewise.
20066 (avx_vpermilp_parallel): Likewise.
20067 (inline_memory_move_cost): Likewise.
20068 (ix86_tieable_integer_mode_p): Likewise.
20069 (x86_maybe_negate_const_int): Likewise.
20070 (ix86_expand_vector_init_duplicate): Likewise.
20071 (ix86_expand_vector_init_one_nonzero): Likewise.
20072 (ix86_expand_vector_init_one_var): Likewise.
20073 (ix86_expand_vector_init_concat): Likewise.
20074 (ix86_expand_vector_init_interleave): Likewise.
20075 (ix86_expand_vector_init_general): Likewise.
20076 (ix86_expand_vector_set): Likewise.
20077 (ix86_expand_vector_extract): Likewise.
20078 (emit_reduc_half): Likewise.
20079 (ix86_emit_i387_round): Likewise.
20080 (ix86_mangle_type): Likewise.
20081 (ix86_expand_round_sse4): Likewise.
20082 (expand_vec_perm_blend): Likewise.
20083 (canonicalize_vector_int_perm): Likewise.
20084 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
20085 (expand_vec_perm_1): Likewise.
20086 (expand_vec_perm_interleave3): Likewise.
20087 (expand_vec_perm_even_odd_pack): Likewise.
20088 (expand_vec_perm_even_odd_1): Likewise.
20089 (expand_vec_perm_broadcast_1): Likewise.
20090 (ix86_vectorize_vec_perm_const_ok): Likewise.
20091 (ix86_expand_vecop_qihi): Likewise.
20092 (ix86_expand_mul_widen_hilo): Likewise.
20093 (ix86_expand_sse2_abs): Likewise.
20094 (ix86_expand_pextr): Likewise.
20095 (ix86_expand_pinsr): Likewise.
20096 (ix86_preferred_simd_mode): Likewise.
20097 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
20098 * config/i386/sse.md (*andnot<mode>3): Likewise.
20099 (<mask_codefor><code><mode>3<mask_name>): Likewise.
20100 (*<code><mode>3): Likewise.
20101 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
20102 (ia64_expand_atomic_op): Likewise.
20103 (ia64_arg_type): Likewise.
20104 (ia64_mode_to_int): Likewise.
20105 (ia64_scalar_mode_supported_p): Likewise.
20106 (ia64_vector_mode_supported_p): Likewise.
20107 (expand_vec_perm_broadcast): Likewise.
20108 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
20109 (iq2000_function_arg_advance): Likewise.
20110 (iq2000_function_arg): Likewise.
20111 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
20112 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
20113 (m68k_libcall_value): Likewise.
20114 (m68k_function_value): Likewise.
20115 (sched_attr_op_type): Likewise.
20116 * config/mcore/mcore.c (mcore_output_move): Likewise.
20117 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
20119 (microblaze_function_arg): Likewise.
20120 * config/mips/mips.c (mips16_build_call_stub): Likewise.
20121 (mips_print_operand): Likewise.
20122 (mips_mode_ok_for_mov_fmt_p): Likewise.
20123 (mips_vector_mode_supported_p): Likewise.
20124 (mips_preferred_simd_mode): Likewise.
20125 (mips_expand_vpc_loongson_even_odd): Likewise.
20126 (mips_expand_vec_unpack): Likewise.
20127 (mips_expand_vi_broadcast): Likewise.
20128 (mips_expand_vector_init): Likewise.
20129 (mips_expand_vec_reduc): Likewise.
20130 (mips_expand_msa_cmp): Likewise.
20131 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
20132 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
20133 (cc_flags_for_mode): Likewise.
20134 * config/msp430/msp430.c (msp430_print_operand): Likewise.
20135 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
20136 (nds32_output_casesi_pc_relative): Likewise.
20137 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20138 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
20139 (nvptx_gen_unpack): Likewise.
20140 (nvptx_gen_pack): Likewise.
20141 (nvptx_gen_shuffle): Likewise.
20142 (nvptx_gen_wcast): Likewise.
20143 (nvptx_preferred_simd_mode): Likewise.
20144 * config/pa/pa.c (pa_secondary_reload): Likewise.
20145 * config/pa/predicates.md (base14_operand): Likewise.
20146 * config/powerpcspe/powerpcspe-c.c
20147 (altivec_resolve_overloaded_builtin): Likewise.
20148 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
20150 (rs6000_preferred_simd_mode): Likewise.
20151 (output_vec_const_move): Likewise.
20152 (rs6000_expand_vector_extract): Likewise.
20153 (rs6000_split_vec_extract_var): Likewise.
20154 (reg_offset_addressing_ok_p): Likewise.
20155 (rs6000_legitimate_offset_address_p): Likewise.
20156 (rs6000_legitimize_address): Likewise.
20157 (rs6000_emit_set_const): Likewise.
20158 (rs6000_const_vec): Likewise.
20159 (rs6000_emit_move): Likewise.
20160 (spe_build_register_parallel): Likewise.
20161 (rs6000_darwin64_record_arg_recurse): Likewise.
20162 (swap_selector_for_mode): Likewise.
20163 (spe_init_builtins): Likewise.
20164 (paired_init_builtins): Likewise.
20165 (altivec_init_builtins): Likewise.
20166 (do_load_for_compare): Likewise.
20167 (rs6000_generate_compare): Likewise.
20168 (rs6000_expand_float128_convert): Likewise.
20169 (emit_load_locked): Likewise.
20170 (emit_store_conditional): Likewise.
20171 (rs6000_output_function_epilogue): Likewise.
20172 (rs6000_handle_altivec_attribute): Likewise.
20173 (rs6000_function_value): Likewise.
20174 (emit_fusion_gpr_load): Likewise.
20175 (emit_fusion_p9_load): Likewise.
20176 (emit_fusion_p9_store): Likewise.
20177 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
20178 (fusion_gpr_mem_load): Likewise.
20179 (fusion_addis_mem_combo_load): Likewise.
20180 (fusion_addis_mem_combo_store): Likewise.
20181 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
20182 (fusion_gpr_mem_load): Likewise.
20183 (fusion_addis_mem_combo_load): Likewise.
20184 (fusion_addis_mem_combo_store): Likewise.
20185 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20187 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
20188 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20189 (rs6000_preferred_simd_mode): Likewise.
20190 (output_vec_const_move): Likewise.
20191 (rs6000_expand_vector_extract): Likewise.
20192 (rs6000_split_vec_extract_var): Likewise.
20193 (reg_offset_addressing_ok_p): Likewise.
20194 (rs6000_legitimate_offset_address_p): Likewise.
20195 (rs6000_legitimize_address): Likewise.
20196 (rs6000_emit_set_const): Likewise.
20197 (rs6000_const_vec): Likewise.
20198 (rs6000_emit_move): Likewise.
20199 (rs6000_darwin64_record_arg_recurse): Likewise.
20200 (swap_selector_for_mode): Likewise.
20201 (paired_init_builtins): Likewise.
20202 (altivec_init_builtins): Likewise.
20203 (rs6000_expand_float128_convert): Likewise.
20204 (emit_load_locked): Likewise.
20205 (emit_store_conditional): Likewise.
20206 (rs6000_output_function_epilogue): Likewise.
20207 (rs6000_handle_altivec_attribute): Likewise.
20208 (rs6000_function_value): Likewise.
20209 (emit_fusion_gpr_load): Likewise.
20210 (emit_fusion_p9_load): Likewise.
20211 (emit_fusion_p9_store): Likewise.
20212 * config/rx/rx.c (rx_gen_move_template): Likewise.
20213 (flags_from_mode): Likewise.
20214 * config/s390/predicates.md (s390_alc_comparison): Likewise.
20215 (s390_slb_comparison): Likewise.
20216 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
20217 (s390_vector_mode_supported_p): Likewise.
20218 (s390_cc_modes_compatible): Likewise.
20219 (s390_match_ccmode_set): Likewise.
20220 (s390_canonicalize_comparison): Likewise.
20221 (s390_emit_compare_and_swap): Likewise.
20222 (s390_branch_condition_mask): Likewise.
20223 (s390_rtx_costs): Likewise.
20224 (s390_secondary_reload): Likewise.
20225 (__SECONDARY_RELOAD_CASE): Likewise.
20226 (s390_expand_cs): Likewise.
20227 (s390_preferred_simd_mode): Likewise.
20228 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
20229 * config/sh/sh.c (sh_print_operand): Likewise.
20230 (dump_table): Likewise.
20231 (sh_secondary_reload): Likewise.
20232 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20233 * config/sh/sh.md (casesi_worker_1): Likewise.
20234 (casesi_worker_2): Likewise.
20235 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
20236 (fcc_comparison_operator): Likewise.
20237 * config/sparc/sparc.c (sparc_expand_move): Likewise.
20238 (emit_soft_tfmode_cvt): Likewise.
20239 (sparc_preferred_simd_mode): Likewise.
20240 (output_cbranch): Likewise.
20241 (sparc_print_operand): Likewise.
20242 (sparc_expand_vec_perm_bmask): Likewise.
20243 (vector_init_bshuffle): Likewise.
20244 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
20245 (spu_vector_mode_supported_p): Likewise.
20246 (spu_expand_insv): Likewise.
20247 (spu_emit_branch_or_set): Likewise.
20248 (spu_handle_vector_attribute): Likewise.
20249 (spu_builtin_splats): Likewise.
20250 (spu_builtin_extract): Likewise.
20251 (spu_builtin_promote): Likewise.
20252 (spu_expand_sign_extend): Likewise.
20253 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
20254 (tilegx_simd_int): Likewise.
20255 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
20256 (tilepro_simd_int): Likewise.
20257 * config/v850/v850.c (const_double_split): Likewise.
20258 (v850_print_operand): Likewise.
20259 (ep_memory_offset): Likewise.
20260 * config/vax/vax.c (vax_rtx_costs): Likewise.
20261 (vax_output_int_move): Likewise.
20262 (vax_output_int_add): Likewise.
20263 (vax_output_int_subtract): Likewise.
20264 * config/visium/predicates.md (visium_branch_operator): Likewise.
20265 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
20266 (visium_print_operand_address): Likewise.
20267 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20268 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
20269 (xtensa_expand_conditional_branch): Likewise.
20270 (xtensa_copy_incoming_a7): Likewise.
20271 (xtensa_output_literal): Likewise.
20272 * dfp.c (decimal_real_maxval): Likewise.
20273 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
20275 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20276 Alan Hayward <alan.hayward@arm.com>
20277 David Sherwood <david.sherwood@arm.com>
20279 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
20280 (mode_nunits_inline): Likewise.
20281 (mode_inner_inline): Likewise.
20282 (mode_unit_size_inline): Likewise.
20283 (mode_unit_precision_inline): Likewise.
20284 (emit_insn_modes_h): Likewise. Also emit a #define of the
20286 (emit_mode_wider): Add an E_ prefix to mode names.
20287 (emit_mode_complex): Likewise.
20288 (emit_mode_inner): Likewise.
20289 (emit_mode_adjustments): Likewise.
20290 (emit_mode_int_n): Likewise.
20291 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
20292 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
20293 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
20294 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
20295 (CRC32_BUILTIN, ENTRY): Likewise.
20296 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
20297 (aarch64_pop_regs): Likewise.
20298 (aarch64_process_components): Likewise.
20299 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
20300 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
20301 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
20302 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
20303 * config/arm/arm.c (arm_init_libfuncs): Likewise.
20304 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
20306 * config/i386/i386-builtin.def (pcmpestr): Likewise.
20307 (pcmpistr): Likewise.
20308 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
20309 * config/mmix/mmix.c (mmix_output_condition): Likewise.
20310 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20312 * config/rl78/rl78.c (mduc_regs): Likewise.
20313 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20314 (htm_expand_builtin): Likewise.
20315 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
20316 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
20317 * config/xtensa/xtensa.c (print_operand): Likewise.
20318 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
20319 (NUM_MODE_VECTOR_INT): Likewise.
20320 * genoutput.c (null_operand): Likewise.
20321 (output_operand_data): Likewise.
20322 * genrecog.c (print_parameter_value): Likewise.
20323 * lra.c (debug_operand_data): Likewise.
20325 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20327 * dwarf2out.c (scompare_loc_descriptor_wide)
20328 (scompare_loc_descriptor_narrow): New functions, split out from...
20329 (scompare_loc_descriptor): ...here.
20330 * expmed.c (emit_store_flag_int): New function, split out from...
20331 (emit_store_flag): ...here.
20333 2017-08-30 Richard Biener <rguenther@suse.de>
20335 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
20336 (dwarf2out_early_finish): Move setting of AT_pubnames from
20337 early debug output to early finish.
20339 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20341 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
20342 and -mdata-region to the assembler.
20344 2017-08-30 Richard Biener <rguenther@suse.de>
20346 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
20348 (gen_subprogram_die): Add DW_AT_object_pointer only early.
20349 (dwarf2out_early_global_decl): Only generate a DIE for the
20350 abstract origin if it doesn't already exist or is a declaration DIE.
20351 (resolve_addr): Do not add the linkage name twice when
20352 generating a stub DIE for the DW_TAG_GNU_call_site target.
20354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20356 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20357 Use machine_mode rather than int for arg1_mode.
20359 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
20362 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
20363 that the second argument of the built-in functions to unpack
20364 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
20365 switch statement instead a lot of if statements.
20366 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
20367 Allow 64-bit values to be in Altivec registers as well as
20368 traditional floating point registers.
20369 (pack<mode>, FMOVE128_VSX iterator): Likewise.
20371 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
20373 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
20374 MAX_REGS_PER_ADDRESS == 1.
20376 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
20378 * config/i386/i386.opt (flag_fentry): Do not init to -1.
20379 * config/i386/i386.c (ix86_option_override_internal): Simplify
20380 setting of opts->x_flag_entry.
20382 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20383 Jakub Jelinek <jakub@redhat.com>
20384 Richard Biener <rguenther@suse.de>
20386 PR tree-optimization/81503
20387 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
20388 folded constant fits in the target type; reorder tests for clarity.
20390 2017-08-29 Martin Liska <mliska@suse.cz>
20392 * passes.def: Include pass_lower_switch.
20393 * stmt.c (dump_case_nodes): Remove and move to
20394 tree-switch-conversion.
20395 (case_values_threshold): Likewise.
20396 (expand_switch_as_decision_tree_p): Likewise.
20397 (emit_case_decision_tree): Likewise.
20398 (expand_case): Likewise.
20399 (balance_case_nodes): Likewise.
20400 (node_has_low_bound): Likewise.
20401 (node_has_high_bound): Likewise.
20402 (node_is_bounded): Likewise.
20403 (emit_case_nodes): Likewise.
20404 (struct simple_case_node): New struct.
20405 (add_case_node): Remove.
20406 (emit_case_dispatch_table): Use vector instead of case_list.
20407 (reset_out_edges_aux): Remove.
20408 (compute_cases_per_edge): Likewise.
20409 (expand_case): Build list of simple_case_node.
20410 (expand_sjlj_dispatch_table): Use it.
20411 * tree-switch-conversion.c (struct case_node): Moved from
20412 stmt.c and adjusted.
20413 (emit_case_nodes): Likewise.
20414 (node_has_low_bound): Likewise.
20415 (node_has_high_bound): Likewise.
20416 (node_is_bounded): Likewise.
20417 (case_values_threshold): Likewise.
20418 (reset_out_edges_aux): Likewise.
20419 (compute_cases_per_edge): Likewise.
20420 (add_case_node): Likewise.
20421 (dump_case_nodes): Likewise.
20422 (balance_case_nodes): Likewise.
20423 (expand_switch_as_decision_tree_p): Likewise.
20424 (emit_jump): Likewise.
20425 (emit_case_decision_tree): Likewise.
20426 (try_switch_expansion): Likewise.
20427 (do_jump_if_equal): Likewise.
20428 (emit_cmp_and_jump_insns): Likewise.
20429 (fix_phi_operands_for_edge): New function.
20430 (record_phi_operand_mapping): Likewise.
20431 (class pass_lower_switch): New pass.
20432 (pass_lower_switch::execute): New function.
20433 (make_pass_lower_switch): Likewise.
20434 (conditional_probability):
20435 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
20436 * tree-pass.h: Add make_pass_lower_switch.
20438 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20441 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
20444 2017-08-29 Richard Biener <rguenther@suse.de>
20446 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
20447 we do not add a DW_AT_inline attribute twice.
20448 (gen_subprogram_die): Remove code setting DW_AT_inline on
20449 DECL_ABSTRACT_P nodes.
20451 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
20453 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
20454 calls to internal functions.
20455 (gimplify_modify_expr): Likewise.
20456 * tree-call-cdce.c (use_internal_fn): Likewise.
20457 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
20458 (convert_to_divmod): Set the nothrow flag.
20459 * tree-if-conv.c (predicate_mem_writes): Likewise.
20460 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20461 (vectorizable_call): Likewise.
20462 (vectorizable_store): Likewise.
20463 (vectorizable_load): Likewise.
20464 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
20465 (vect_recog_mask_conversion_pattern): Likewise.
20467 2017-08-29 Martin Liska <mliska@suse.cz>
20470 * gcc.c (driver_handle_option): Add new argument.
20471 * opts-common.c (handle_option): Pass
20472 target_option_override_hook.
20473 * opts-global.c (lang_handle_option): Add new option.
20474 (set_default_handlers): Add new argument.
20475 (decode_options): Likewise.
20476 * opts.c (target_handle_option): Likewise.
20477 (common_handle_option): Call target_option_override_hook.
20478 * opts.h (struct cl_option_handler_func): Add hook for
20479 target option override.
20480 (struct cl_option_handlers): Likewise.
20481 (set_default_handlers): Add new argument.
20482 (decode_options): Likewise.
20483 (common_handle_option): Likewise.
20484 (target_handle_option): Likewise.
20485 * toplev.c (toplev::main): Pass targetm.target_option.override
20488 2017-08-29 Richard Biener <rguenther@suse.de>
20489 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
20491 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
20492 life to the active subtree.
20494 2017-08-28 Jeff Law <law@redhat.com>
20496 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
20497 derive_equivalences.
20498 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
20499 Code moved into....
20500 (edge_info::derive_equivalences): New private member function
20502 * tree-ssa-dom.c (class edge_info): Changed from a struct
20503 to a class. Add ctor/dtor, methods and data members.
20504 (edge_info::edge_info): Renamed from allocate_edge_info.
20505 Initialize additional members.
20506 (edge_info::~edge_info): New.
20507 (free_dom_edge_info): Delete the edge info.
20508 (record_edge_info): Use new class & associated member functions.
20509 Tighten forms for testing for edge equivalences.
20510 (record_temporary_equivalences): Iterate over the simple
20511 equivalences rather than assuming there's only one per edge.
20512 (cprop_into_successor_phis): Iterate over the simple
20513 equivalences rather than assuming there's only one per edge.
20514 (optimize_stmt): Use operand_equal_p rather than pointer
20515 equality for mini-DSE code.
20517 2017-08-28 Nathan Sidwell <nathan@acm.org>
20519 * gcc.c (execute): Fold SIGPIPE handling into switch
20520 statement. Adjust internal error message.
20522 2017-08-28 Richard Biener <rguenther@suse.de>
20525 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
20526 Do nothing for removed DIEs.
20528 2017-08-28 Richard Biener <rguenther@suse.de>
20530 PR tree-optimization/81977
20531 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
20534 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
20537 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
20538 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
20539 using targetm.gen_mem_thread_fence.
20541 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
20544 * config/i386/i386.md (*<btsc><mode>): Change operand 2
20545 predicate to register_operand. Reorder operands.
20546 (*btr<mode>): Ditto.
20547 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
20548 (*btr<mode>_mask): Ditto.
20550 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
20552 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
20553 * config/rs6000/xmmintrin.h: New file.
20554 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
20556 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20559 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
20560 parameter and_insn and return it.
20561 (recombine_lvx_pattern): Insert a copy to ensure availability of
20562 the base register of the copied masking operation at the point of
20563 the instruction replacement.
20564 (recombine_stvx_pattern): Likewise.
20566 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20568 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
20569 undocumented switches.
20570 (-mpower9-dform-vector): Likewise.
20571 (-mpower9-dform): Likewise.
20572 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
20573 comments to delete references to -mpower9-dform* switches.
20574 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
20575 Delete reference to -mpower9-dform* switches, test for
20576 -mpower9-vector instead.
20577 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
20578 (OTHER_P9_VECTOR_MASKS): Likewise.
20579 (POWERPC_MASKS): Likewise.
20580 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
20581 tests against -mpower9-dform* to -mpower9-vector. Delete code
20582 that checked for -mpower9-dform* consistancy with other options.
20583 Add test for -mpower9-misc to enable other power9 switches.
20584 (rs6000_init_hard_regno_mode_ok): Likewise.
20585 (rs6000_option_override_internal): Likewise.
20586 (rs6000_emit_prologue): Likewise.
20587 (rs6000_emit_epilogue): Likewise.
20588 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
20589 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
20590 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
20592 (emit_fusion_p9_store): Likewise.
20593 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
20594 resetting these macros if the assembler does not support ISA 3.0
20596 (TARGET_P9_DFORM_VECTOR): Likewise.
20597 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
20598 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
20600 2017-08-25 Alan Modra <amodra@gmail.com>
20602 PR rtl-optimization/81747
20603 * cse.c (cse_extended_basic_block): Don't attempt to record
20604 equivalences for degenerate conditional branches that branch
20605 to their fall-through.
20607 2017-08-24 Martin Sebor <msebor@redhat.com>
20609 PR middle-end/81908
20610 * gimple-fold.c (size_must_be_zero_p): New function.
20611 (gimple_fold_builtin_memory_op): Call it.
20613 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
20615 * config/rs6000/mm_malloc.h: New file.
20617 2017-08-24 Bin Cheng <bin.cheng@arm.com>
20619 PR tree-optimization/81913
20620 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
20621 analysis when either IVs in condition can wrap.
20623 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
20625 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
20626 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
20628 2017-08-24 Richard Biener <rguenther@suse.de>
20631 * targhooks.c (default_target_can_inline_p): Properly
20632 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
20633 is present and always compare.
20634 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
20635 infer -mfpmath=sse from TARGET_SSE_P.
20636 (ix86_can_inline_p): Properly use target_option_default_node when
20637 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
20639 2017-08-24 Richard Biener <rguenther@suse.de>
20642 * dwarf2out.c (output_die): Handle flag_generate_offload like
20644 (output_comp_unit): Likewise.
20645 (gen_array_type_die): Likewise.
20646 (dwarf2out_early_finish): Likewise.
20647 (note_variable_value_in_expr): Likewise.
20648 (dwarf2out_finish): Likewise. Adjust assert.
20649 * cgraphunit.c (symbol_table::compile): Move setting of
20650 flag_generate_offload earlier ...
20651 (symbol_table::finalize_compilation_unit): ... here, before
20652 early debug finalization.
20654 2017-08-24 Richard Biener <rguenther@suse.de>
20656 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
20657 and ipa-fnsummary.h.
20658 (ix86_can_inline_p): When ix86_fpmath flags do not match
20659 check whether the callee uses FP math at all.
20661 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
20663 PR middle-end/81931
20664 * tree-ssanames.c (get_nonzero_bits): Use element_precision
20665 instead of TYPE_PRECISION.
20667 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
20668 Alan Hayward <alan.hayward@arm.com>
20669 David Sherwood <david.sherwood@arm.com>
20671 * combine.c (make_extraction): Use subreg_offset_from_lsb.
20673 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
20675 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
20677 (ix86_frame::stack_realign_allocate): New field.
20678 (struct machine_frame_state): Modify comments.
20679 (machine_frame_state::sp_realigned_fp_end): New field.
20680 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
20681 layout calculation.
20682 (sp_valid_at): Add assertion to assure no attempt to access invalid
20683 offset of a realigned stack.
20684 (fp_valid_at): Likewise.
20685 (choose_baseaddr): Modify comments.
20686 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
20687 ix86_expand_prologue.
20688 (ix86_expand_prologue): Modify stack realignment and allocation.
20689 (ix86_expand_epilogue): Modify comments.
20690 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
20691 avx2_runtime, avx512f, and avx512f_runtime.
20693 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
20695 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
20696 (mstackrealign): Do not init to -1.
20697 * config/i386/i386.c (ix86_option_override_internal):
20698 Check opts_set, not opts when setting default value of
20699 opts->x_ix86_force_align_arg_pointer.
20701 2017-08-23 Richard Biener <rguenther@suse.de>
20703 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
20704 lang_hooks.decl_printable_name.
20705 * print-rtl-function.c (print_rtx_function): Likewise.
20706 * tree-pretty-print.c (dump_function_header): Likewise.
20708 2017-08-23 Richard Biener <rguenther@suse.de>
20711 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
20712 -g0 at compile-time.
20714 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20716 PR middle-end/19706
20717 * doc/sourcebuild.texi (Other hardware attributes):
20720 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20722 PR middle-end/19706
20723 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
20724 Fix single-use check.
20726 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
20728 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
20730 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20732 * doc/install.texi: Modify to add more details on running selected
20735 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20737 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
20738 is combined with -mabi=ms.
20739 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
20742 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20744 PR tree-optimization/81488
20745 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
20746 and cached_basis fields.
20747 (MAX_SPREAD): New constant.
20748 (alloc_cand_and_find_basis): Initialize new fields.
20749 (clear_visited): New function.
20750 (create_phi_basis_1): Rename from create_phi_basis, set visited
20751 and cached_basis fields.
20752 (create_phi_basis): New wrapper function.
20753 (phi_add_costs_1): Rename from phi_add_costs, add spread
20754 parameter, set visited field, short-circuit when limits reached.
20755 (phi_add_costs): New wrapper function.
20756 (record_phi_increments_1): Rename from record_phi_increments, set
20758 (record_phi_increments): New wrapper function.
20759 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
20760 (phi_incr_cost): New wrapper function.
20761 (all_phi_incrs_profitable_1): Rename from
20762 all_phi_incrs_profitable, set visited field.
20763 (all_phi_incrs_profitable): New wrapper function.
20765 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20766 Alan Hayward <alan.hayward@arm.com>
20767 David Sherwood <david.sherwood@arm.com>
20769 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
20770 that takes the outer and inner modes.
20771 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
20772 comparison as the canonical test for a paradoxical subreg.
20773 * combine.c (simplify_set): Use paradoxical_subreg_p.
20774 (make_extraction): Likewise.
20775 (force_to_mode): Likewise.
20776 (rtx_equal_for_field_assignment_p): Likewise.
20777 (gen_lowpart_for_combine): Likewise.
20778 (simplify_comparison): Likewise.
20779 * cse.c (equiv_constant): Likewise.
20780 * expmed.c (store_bit_field_1): Likewise.
20781 * final.c (alter_subreg): Likewise.
20782 * fwprop.c (propagate_rtx): Likewise.
20783 (forward_propagate_subreg): Likewise.
20784 * ira-conflicts.c (ira_build_conflicts): Likewise.
20785 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
20786 * lra-constraints.c (curr_insn_transform): Likewise.
20787 (split_reg): Likewise.
20788 * lra-eliminations.c (move_plus_up): Likewise.
20789 (lra_eliminate_regs_1): Likewise.
20790 * recog.c (general_operand): Likewise.
20791 * ree.c (combine_reaching_defs): Likewise.
20792 * reload.c (push_reload): Likewise.
20793 (find_reloads): Likewise.
20794 * reload1.c (elimination_effects): Likewise.
20795 (compute_reload_subreg_offset): Likewise.
20796 (choose_reload_regs): Likewise.
20797 * rtlanal.c (subreg_lsb_1): Likewise.
20798 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
20799 (simplify_subreg): Likewise.
20800 * var-tracking.c (track_loc_p): Likewise.
20801 * emit-rtl.c (byte_lowpart_offset): Likewise.
20802 (paradoxical_subreg_p): Delete out-of-line definition.
20804 2017-08-22 Jeff Law <law@redhat.com>
20806 PR tree-optimization/81741
20807 PR tree-optimization/71947
20808 * tree-ssa-dom.c: Include tree-inline.h.
20809 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20810 equivalences if one is more expensive to compute than the other.
20811 * tree-ssa-scopedtables.h (class const_or_copies): Make
20812 record_const_or_copy_raw method private.
20813 (class avail_exprs_stack): New method simplify_binary_operation.
20814 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20815 avail_exprs_stack::simplify_binary_operation as needed.
20816 (avail_exprs_stack::simplify_binary_operation): New function.
20818 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20820 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20821 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20822 (DOT_SYMBOLS): Likewise.
20823 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20824 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20825 (RS6000_ABI_NAME): Likewise.
20826 (TARGET_CMODEL): Likewise.
20827 (TOC_SECTION_ASM_OP): Likewise.
20828 (SET_CMODEL): New macro.
20829 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20831 2017-08-22 Richard Biener <rguenther@suse.de>
20833 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20834 to allow for free-lang-data replacements similar to verify_type_variant.
20836 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20837 Alan Hayward <alan.hayward@arm.com>
20838 David Sherwood <david.sherwood@arm.com>
20840 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20841 VOIDmode for the LABEL_REF.
20843 2017-08-22 Richard Biener <rguenther@suse.de>
20845 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20847 2017-08-22 Richard Biener <rguenther@suse.de>
20849 * common.opt (feliminate-dwarf2-dups): Ignore.
20850 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20851 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20852 same_die_p_wrap, compute_section_prefix,
20853 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20854 (comdat_symbol_id, comdat_symbol_number): Likewise.
20855 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20857 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20858 (output_die): Mark unreachable path unreachable.
20859 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20860 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20861 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20862 (dwarf2out_early_finish): Likewise.
20864 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20866 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20868 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20871 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20872 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20873 (avr_attribute_table) <io, io_low, address>: Initialize
20874 .decl_required with true.
20876 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20878 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20879 undocumented debugging options.
20880 (-mvsx-scalar-double): Likewise.
20881 (-mallow-df-permute): Likewise.
20882 (-mvectorize-builtins): Likewise.
20883 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20884 (rs6000_builtin_vectorized_function): Likewise.
20885 (rs6000_builtin_md_vectorized_function): Likewise.
20886 (rs6000_opt_vars): Likewise.
20888 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20891 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20892 (*btrq_imm): Rename from *btrq.
20893 (*btcq_imm): Rename from *btcq.
20894 (btsc): New code attribute.
20895 (*<btsc><mode>): New insn pattern.
20896 (*btr<mode>): Ditto.
20897 (*<btsc><mode>_mask): New insn_and_split pattern.
20898 (*btr<mode>_mask): Ditto.
20900 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20901 Alan Hayward <alan.hayward@arm.com>
20902 David Sherwood <david.sherwood@arm.com>
20904 * function.c (pad_below): Simplify padding calculation.
20906 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20907 Alan Hayward <alan.hayward@arm.com>
20908 David Sherwood <david.sherwood@arm.com>
20910 * target.def (function_prologue): Remove frame size argument.
20911 (function_epilogue): Likewise.
20912 * doc/tm.texi: Regenerate.
20913 * final.c (final_start_function): Update call to function_prologue.
20914 (final_end_function): Update call to function_epilogue.
20915 (default_function_pro_epilogue): Remove frame size argument.
20916 * output.h (default_function_pro_epilogue): Likewise.
20917 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
20918 (arm_output_function_prologue): Likewise.
20919 * config/frv/frv.c (frv_function_prologue): Likewise.
20920 (frv_function_epilogue): Likewise.
20921 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
20922 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
20923 (ia64_output_function_epilogue): Likewise.
20924 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
20925 (m32r_output_function_epilogue): Likewise.
20926 * config/microblaze/microblaze.c (microblaze_function_prologue)
20927 (microblaze_function_epilogue): Likewise.
20928 * config/mips/mips.c (mips_output_function_prologue): Likewise.
20929 (mips_output_function_epilogue): Likewise.
20930 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
20931 (mmix_target_asm_function_epilogue): Likewise.
20932 * config/msp430/msp430.c (msp430_start_function): Likewise.
20933 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
20934 (nds32_asm_function_epilogue): Likewise.
20935 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
20936 * config/pa/pa.c (pa_output_function_prologue): Likewise.
20937 (pa_output_function_epilogue): Likewise.
20938 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
20939 (rs6000_output_function_epilogue): Likewise.
20940 * config/rl78/rl78.c (rl78_start_function): Likewise.
20941 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
20942 (rs6000_output_function_epilogue): Likewise.
20943 * config/rx/rx.c (rx_output_function_prologue): Likewise.
20944 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
20945 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
20946 (sparc_asm_function_epilogue): Likewise.
20948 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20950 * tree.h (type_has_mode_precision_p): New function.
20951 * convert.c (convert_to_integer_1): Use it.
20952 * expr.c (expand_expr_real_2): Likewise.
20953 (expand_expr_real_1): Likewise.
20954 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
20955 * match.pd: Likewise.
20956 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
20957 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
20958 * tree-tailcall.c (process_assignment): Likewise.
20959 * tree-vect-loop.c (vectorizable_reduction): Likewise.
20960 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
20961 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
20962 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
20963 (vectorizable_assignment): Likewise.
20964 (vectorizable_shift): Likewise.
20965 (vectorizable_operation): Likewise.
20966 * tree-vrp.c (register_edge_assert_for_2): Likewise.
20968 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
20970 * match.pd: Add pow (C, x) simplification.
20972 2017-08-21 Richard Biener <rguenther@suse.de>
20974 PR tree-optimization/81900
20975 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
20976 for blocks with abnormal predecessors.
20977 (compute_antic): Do not set visited flag prematurely.
20979 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
20982 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
20984 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20986 * stor-layout.h (vector_type_mode): Move to...
20987 * tree.h (vector_type_mode): ...here.
20988 * stor-layout.c (vector_type_mode): Move to...
20989 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
20991 2017-08-21 Richard Biener <rguenther@suse.de>
20993 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
20994 register_external_die hooks.
20995 (debug_false_tree_charstarstar_uhwistar): Declare.
20996 (debug_nothing_tree_charstar_uhwi): Likewise.
20997 * debug.c (do_nothing_debug_hooks): Adjust.
20998 (debug_false_tree_charstarstar_uhwistar): New do nothing.
20999 (debug_nothing_tree_charstar_uhwi): Likewise.
21000 * dbxout.c (dbx_debug_hooks): Adjust.
21001 (xcoff_debug_hooks): Likewise.
21002 * sdbout.c (sdb_debug_hooks): Likewise.
21003 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21004 * dwarf2out.c (macinfo_label_base): New global.
21005 (dwarf2out_register_external_die): New function for the
21006 register_external_die hook.
21007 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
21008 (dwarf2_debug_hooks): Use them.
21009 (dwarf2_lineno_debug_hooks): Adjust.
21010 (struct die_struct): Add with_offset flag.
21011 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
21012 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
21013 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
21014 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
21015 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
21016 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
21017 defining section names for the early LTO debug variants.
21018 (reset_indirect_string): New helper.
21019 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
21020 (print_dw_val): Add support for offsetted symbol references.
21021 (get_ultimate_context): Split out from is_cxx.
21022 (is_cxx): Use get_ultimate_context.
21023 (is_fortran): Add decl overload.
21024 (compute_comp_unit_symbol): Split out worker from
21025 compute_section_prefix.
21026 (compute_section_prefix): Call compute_comp_unit_symbol and
21027 set comdat_type_p here.
21028 (output_die): Skip DIE symbol output for the LTO added one.
21029 Handle DIE symbol references with offset.
21030 (output_comp_unit): Guard section name mangling properly.
21031 For LTO debug sections emit a symbol at the section beginning
21032 which we use to refer to its DIEs.
21033 (add_abstract_origin_attribute): For DIEs registered via
21034 dwarf2out_register_external_die directly refer to the early
21035 DIE rather than indirectly through the shadow one we created.
21036 Remove obsolete call to dwarf2out_abstract_function for
21037 non-function/block origins.
21038 (gen_array_type_die): When generating early LTO debug do
21039 not emit DW_AT_string_length.
21040 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
21041 late when in LTO. As suggested place a gcc_unreachable for
21042 the DECL_ABSTRACT_P case.
21043 (gen_subprogram_die): Avoid another specification DIE
21044 for early built declarations/definitions for the late LTO case.
21045 (gen_variable_die): Add type references for late duplicated VLA dies
21047 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
21048 we have the abstract instance already.
21049 (process_scope_var): Adjust decl DIE contexts in LTO which
21050 first puts them in limbo.
21051 (gen_decl_die): Do not generate type DIEs late apart from
21052 types for VLAs or for decls we do not yet have a DIE. Do not
21053 call dwarf2out_abstract_function late.
21054 (dwarf2out_early_global_decl): Make sure to create DIEs
21055 for abstract instances of a decl first.
21056 (dwarf2out_late_global_decl): Adjust comment.
21057 (output_macinfo_op): With multiple macro sections use
21058 macinfo_label_base to distinguish labels.
21059 (output_macinfo): Likewise. Update macinfo_label_base.
21060 Pass in the line info label.
21061 (note_variable_value_in_expr): When generating LTO resolve
21062 all variable values here by generating DIEs as needed.
21063 (init_sections_and_labels): Add early LTO debug flag parameter
21064 and generate different sections and names if set. Add generation
21065 counter for the labels so we can have multiple of them.
21066 (reset_dies): Helper to allow DIEs to be output multiple times.
21067 (dwarf2out_finish): When outputting DIEs to the fat part of an
21068 LTO object first reset DIEs.
21069 (dwarf2out_early_finish): Output early DIEs when generating LTO.
21070 (modified_type_die): Check for decl_ultimate_origin being self
21072 (gen_type_die_with_usage): Likewise.
21073 (gen_typedef_die): Allow decl_ultimate_origin being self.
21074 (set_decl_abstract_flags): Remove.
21075 (set_block_abstract_flags): Likewise.
21076 (dwarf2out_abstract_function): Treat the early generated DIEs
21077 as the abstract copy and only add DW_AT_inline and
21078 DW_AT_artificial here and call set_decl_origin_self.
21079 If the DIE has an abstract origin don't do anything.
21080 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
21081 if we have none yet (Go fails to build one, PR78628).
21082 (variably_modified_type_p): Prevent endless recursion for Ada
21083 cyclic pointer types.
21084 * lto-streamer-in.c: Include debug.h.
21085 (dref_queue): New global.
21086 (lto_read_tree_1): Stream in DIE references.
21087 (lto_input_tree): Register DIE references.
21088 (input_function): Stream DECL_DEBUG_ARGS.
21089 * lto-streamer-out.c: Include debug.h.
21090 (lto_write_tree_1): Output DIE references.
21091 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
21092 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
21093 (output_function): Stream DECL_DEBUG_ARGS.
21094 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
21095 Stream DECL_ABSTRACT_ORIGIN.
21096 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
21097 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
21098 DECL_CONTEXT for file-scope decls.
21099 * lto-streamer.h (struct dref_entry): Declare.
21100 (dref_queue): Likewise.
21101 * cfgexpand.c (pass_expand::execute): Do not call the
21102 outlining_inline_function hook here.
21103 * lto-wrapper.c (debug_obj): New global.
21104 (tool_cleanup): Unlink it if required.
21105 (debug_objcopy): New function.
21106 (run_gcc): Handle early debug sections in the IL files by
21107 extracting them to separate files, partially linkin them and
21108 feeding the result back as result to the linker.
21109 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
21110 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
21111 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
21112 sections into a separate segment.
21113 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
21115 (darwin_asm_dwarf_section): Likewise.
21116 (darwin_asm_output_dwarf_offset): Likewise.
21117 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
21119 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21120 Alan Hayward <alan.hayward@arm.com>
21121 David Sherwood <david.sherwood@arm.com>
21123 * read-md.h (md_reader::record_potential_iterator_use): Replace
21124 pointer argument with an rtx and an index.
21125 * read-rtl.c (iterator_group::apply_iterator): Likewise.
21126 (apply_mode_iterator): Likewise.
21127 (apply_code_iterator): Likewise.
21128 (apply_int_iterator): Likewise.
21129 (apply_subst_iterator): Likewise.
21130 (record_iterator_use): Likewise.
21131 (record_attribute_use): Likewise.
21132 (md_reader::record_potential_iterator_use): Likewise. Update calls
21133 to record_iterator_use and apply_iterator.
21134 (iterator_use): Replace ptr with x and index.
21135 (attribute_use): Likewise.
21136 (apply_attribute_uses): Update calls to apply_iterator.
21137 (apply_iterators): Likewise. Update initialization of iterator_use.
21138 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
21139 and record_potential_iterator_use.
21140 (rtx_reader::read_rtx_operand): Likewise.
21142 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21143 Alan Hayward <alan.hayward@arm.com>
21144 David Sherwood <david.sherwood@arm.com>
21146 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
21149 2017-08-21 Richard Biener <rguenther@suse.de>
21151 PR middle-end/81884
21152 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
21153 at struct end conservatively when comparing common bases.
21155 2017-08-21 Richard Biener <rguenther@suse.de>
21157 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
21158 (mem_ref_in_stmt): Remove.
21159 (determine_max_movement): Use ref index to get at the reference.
21160 (invariantness_dom_walker::before_dom_children): Deal with
21161 lim data already initialized.
21162 (gather_mem_refs_stmt): Initialize lim data and record ref index.
21164 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21166 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
21167 (TARGET_ISA_ROUND): Ditto.
21168 (TARGET_ROUND): Ditto.
21169 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
21170 * config/i386/i386.md: Ditto.
21171 * config/i386/sse.md: Ditto.
21172 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
21173 with OPTION_MASK_ISA_SSE4_1.
21175 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21178 * doc/extend.texi (x86 Built-in Functions): Correct the name of
21179 __builtin_ia32_lzcnt_u16.
21181 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
21184 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
21185 (rs6000_set_current_function): Rewrite function to use it.
21187 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
21190 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
21191 and TYPE_WARN_IF_NOT_ALIGN.
21192 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
21193 (handle_warn_if_not_align): New.
21194 (place_union_field): Call handle_warn_if_not_align.
21195 (place_field): Call handle_warn_if_not_align.
21196 Copy TYPE_WARN_IF_NOT_ALIGN.
21197 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
21198 (layout_type): Likewise.
21199 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
21201 (tree_decl_common): Add warn_if_not_align.
21202 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
21203 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
21204 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
21205 (DECL_WARN_IF_NOT_ALIGN): Likewise.
21206 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
21207 * doc/extend.texi: Document warn_if_not_aligned attribute.
21208 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
21210 2017-08-17 Martin Liska <mliska@suse.cz>
21213 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
21214 (get_data_dependence): Use it as pointer type.
21215 (distribute_loop): Likewise.
21217 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21219 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
21220 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
21221 (p8_vmrgow_<mode>_direct): New define_insn.
21222 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
21223 handle endianness for vmrgew and vmrgow permute patterns.
21225 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21227 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
21228 * config/rs6000/rs6000-cpus.def: Remove comment.
21229 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
21230 (POWERPC_MASKS): Likewise.
21231 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
21232 use of TARGET_VSX_TIMODE.
21233 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
21234 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
21235 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
21236 (rs6000_option_override_internal): Remove dead code.
21237 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
21238 (rs6000_legitimize_reload_address): Likewise.
21239 (rs6000_legitimate_address_p): Likewise.
21240 (rs6000_opt_masks): Delete "vsx-timode".
21241 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
21242 from function comment.
21243 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
21244 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
21245 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
21247 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
21248 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
21249 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
21250 (VSX_TI): Likewise.
21252 (define_peephole2): Likewise.
21254 2017-08-17 Martin Sebor <msebor@redhat.com>
21257 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
21258 past the end of an array.
21259 (test_pp_format): Add test cases.
21261 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
21263 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
21264 missing ECF_NOTHROW flags.
21266 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21269 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
21270 operands residing in integer registers.
21271 (*vsx_le_perm_load_<mode>): Likewise.
21272 (*vsx_le_perm_store_<mode>): Likewise.
21273 (define_peephole2): Add peepholes to optimize the above.
21275 2017-08-17 Marek Polacek <polacek@redhat.com>
21277 PR middle-end/81814
21278 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
21279 to mimic what shorten_compare did. Change the return type to bool.
21280 (fold_cond_expr_with_comparison): Update call to
21281 operand_equal_for_comparison_p.
21282 (fold_ternary_loc): Likewise.
21284 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
21286 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
21288 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
21290 2017-08-17 Richard Biener <rguenther@suse.de>
21292 * tree-ssa-structalias.c (solve_graph): When propagating
21293 to successors update the graphs succ edges and avoid duplicate work.
21295 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
21298 * config/i386/i386.c (ix86_option_override_internal): Save target
21299 specific options after ix86_stack_protector_guard_reg was changed.
21301 2017-08-17 Richard Biener <rguenther@suse.de>
21303 PR tree-optimization/81827
21304 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
21305 (new_var_info): Initialize it conservatively.
21306 (get_call_vi): Mark register vars.
21307 (new_scalar_tmp_constraint_exp): Likewise.
21308 (handle_rhs_call): Likewise.
21309 (handle_const_call): Likewise.
21310 (create_function_info_for): Likewise.
21311 (solve_constraints): Sort varinfos to separate register from
21312 non-register vars to pack points-to solution bitmaps during
21315 2017-08-17 Marek Polacek <polacek@redhat.com>
21317 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
21319 2017-08-17 Richard Biener <rguenther@suse.de>
21321 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
21322 to true when overflow is undefined and we saturated the result.
21324 2017-08-17 Alan Modra <amodra@gmail.com>
21327 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
21328 Don't use store multiple if only one reg needs saving.
21329 (interesting_frame_related_regno): New function.
21330 (rs6000_frame_related): Don't emit frame info for regs that
21332 (rs6000_emit_epilogue): Likewise.
21334 2017-08-16 Nathan Sidwell <nathan@acm.org>
21336 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
21337 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
21338 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
21339 * tree.c (free_lang_data_in_type): Use else-if chain. Always
21340 clear TYPE_LANG_1. Remove obsolete member-function stripping.
21341 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
21342 (verify_type): Adjust for TYPE_BINFO move.
21343 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
21344 process TYPE_BINFO directly.
21345 (hash_tree): Likewise.
21346 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21348 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21351 2017-08-16 David Malcolm <dmalcolm@redhat.com>
21353 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
21355 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
21358 * config/i386/i386.md (*anddi_1_btr): Change predicates of
21359 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
21360 Add ix86_binary_operator_ok to insn constraint.
21361 (*iordi_1_bts): Ditto.
21362 (*xordi_1_btc): Ditto.
21363 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
21364 Update corresponding peephole2 pattern.
21368 2017-08-16 Bin Cheng <bin.cheng@arm.com>
21370 PR tree-optimization/81832
21371 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
21372 copy loop header which has IFN_LOOP_DIST_ALIAS call.
21374 2017-08-16 Marek Polacek <polacek@redhat.com>
21377 * fold-const.c (fold_indirect_ref_1): Restore original behavior
21378 regarding size_zero_node.
21380 2017-08-16 Martin Liska <mliska@suse.cz>
21383 * config.gcc: Respect previously set extra_objs in case
21386 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
21388 PR tree-optimization/81835
21389 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
21390 the phi SSA_NAME. Check that the condition in a COND_EXPR does
21391 not depend on the phi.
21393 2017-08-16 Alan Modra <amodra@gmail.com>
21395 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
21398 2017-08-16 Alan Modra <amodra@gmail.com>
21400 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
21401 (save_reg_p): ..into this. Update all callers.
21402 (first_reg_to_save): Simplify.
21404 2017-08-16 Alan Modra <amodra@gmail.com>
21406 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
21409 2017-08-15 Joseph Myers <joseph@codesourcery.com>
21413 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
21414 constant count if that count is less than 32.
21416 2017-08-15 Nathan Sidwell <nathan@acm.org>
21418 * gcc.c (execute): Emit friendlier message if inferior is killed
21419 by an external cause.
21421 2017-08-15 Richard Biener <rguenther@suse.de>
21423 PR tree-optimization/81790
21424 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
21425 CONSTRUCTORs from simplifying and VN.
21427 2017-08-14 Martin Sebor <msebor@redhat.com>
21429 * builtin-attrs.def: Add comments.
21431 2017-08-14 Martin Sebor <msebor@redhat.com>
21434 * doc/extend.texi (attribute nonstring): Document new attribute.
21436 2017-08-14 Martin Sebor <msebor@redhat.com>
21439 * tree-diagnostic.c (default_tree_printer): Handle %G.
21440 * gimple-pretty-print.h (percent_G_format): Declare new function.
21441 * gimple-pretty-print.c (percent_G_format): Define.
21442 * tree-pretty-print.c (percent_K_format): Add argument.
21444 2017-08-14 Martin Sebor <msebor@redhat.com>
21446 PR translation/79998
21447 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
21448 Remove a stray space.
21450 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
21453 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
21454 (*iordi_1_bts): Ditto.
21455 (*xordi_1_btc): Ditto.
21457 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21460 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
21461 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21463 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
21464 quoted strings, and make more translator-friendly.
21465 (darwin_rs6000_override_options): Likewise.
21466 (rs6000_option_override_internal): Likewise.
21467 (rs6000_return_in_memory): Fix overlong line.
21468 (init_cmulative_args): Use quoted strings, and make more
21469 translator-friendly.
21470 (rs6000_pass_by_reference): Fix overlong line.
21471 (def_builtin): Use quoted strings.
21472 (altivec_expand_predicate_builtin): Use quoted strings, and make
21473 more translator-friendly.
21474 (htm_expand_builtin): Use quoted strings.
21475 (cpu_expand_builtin): Use quoted strings, and make more
21476 translator-friendly.
21477 (altivec_expand_builtin): Likewise.
21478 (paired_expand_predicate_builtin): Likewise.
21479 (rs6000_invalid_builtin): Likewise.
21480 (builtin_function_type): Use quoted strings.
21481 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
21482 more translator-friendly.
21483 (rs6000_trampoline_init): Likewise.
21484 (rs6000_handle_altivec_attribute): Likewise.
21485 (rs6000_inner_target_options): Use quoted strings.
21486 (rs6000_disable_incompatible_switches): Likewise.
21487 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
21488 strings, and make more translator-friendly.
21489 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
21491 2017-08-14 Bin Cheng <bin.cheng@arm.com>
21493 PR tree-optimization/81799
21494 * tree-loop-distribution.c (version_loop_by_alias_check): Force
21495 cond_expr to simple gimple operand.
21497 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
21499 PR middle-end/46932
21500 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
21502 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
21506 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
21507 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
21508 TARGET_GASISR_PROLOGUES.
21509 * config/avr/avr.c (avr_option_override): Same.
21510 (avr_pass_pre_proep::execute): Same.
21512 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
21515 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
21516 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
21518 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
21520 * config/i386/i386.md (*load_tp_<mode>): Redefine as
21521 define_insn_and_split. Split to a memory load from 0 in
21522 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
21523 using PTR mode iterator.
21524 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
21525 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
21526 (*add_tp_<mode>): Redefine as define_insn_and_split.
21527 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
21528 address space. Merge with *add_tp_x32 using PTR mode iterator.
21529 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
21530 Split to an add with a memory load from 0 in
21531 DEFAULT_TLS_SEG_REG address space.
21533 2017-08-12 Andrew Pinski <apinski@cavium.com>
21535 * config/aarch64/aarch64-option-extensions.def (rdma):
21536 Fix feature string to what Linux prints out in /proc/cpuinfo.
21538 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
21541 * dwarf2out.c (modified_type_die): For C typedef types that have
21542 an ultimate origin, process the ultimate origin instead of the
21544 (gen_typedef_die): Assert that input DECLs have no ultimate
21546 (gen_type_die_with_usage): For typedef variants that have an
21547 ultimate origin, just call gen_decl_die on the original DECL.
21548 (process_scope_var): Avoid creating DIEs for local typedefs and
21549 concrete static variables.
21551 2017-08-12 Alan Modra <amodra@gmail.com>
21555 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
21556 match gnu-user.h startfile.
21557 (ENDFILE_LINUX_SPEC): Similarly.
21559 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
21562 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
21564 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
21566 2017-08-11 Tamar Christina <tamar.christina@arm.com>
21567 * config/aarch64/aarch64.md (mov<mode>): Change.
21568 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21569 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
21570 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
21572 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
21574 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
21575 for storage order barriers.
21577 2017-08-11 Martin Liska <mliska@suse.cz>
21580 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
21581 variables of void type.
21583 2017-08-11 Martin Liska <mliska@suse.cz>
21585 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
21586 TARGET_SUPPORTS_ALIASES.
21587 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
21588 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
21589 (optimize_weakref): Likewise.
21590 * symtab.c (symtab_node::noninterposable_alias): Likewise.
21591 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
21592 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
21594 2017-08-11 Martin Liska <mliska@suse.cz>
21597 * config/i386/i386.c (make_resolver_func): Do complete
21598 refactoring of the function.
21600 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21603 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
21604 * config/i386/i386.c (ix86_stack_protect_guard): Use
21605 ix86_stack_protect_guard_symbol_str to generate varible declaration.
21606 * doc/invoke.texi (x86 Options): Document
21607 -mstack-protector-guard-symbol= option.
21609 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21611 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
21612 * config/i386/i386.c (ix86_split_stack_guard): New function.
21613 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
21614 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
21615 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
21616 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
21617 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
21618 (split_stack_space_check): Call ix86_split_stack_guard.
21620 2017-08-10 Martin Sebor <msebor@redhat.com>
21622 * print-tree.c (print_node): Print location using the established
21624 Replace spaces with colons.
21625 (debug_raw, debug): Ditto.
21627 2017-08-10 Martin Sebor <msebor@redhat.com>
21630 * pretty-print.c (pp_format): Correct the handling of %s precision.
21632 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
21635 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21637 (ix86_finalize_stack_frame_flags): This. Also clear
21638 frame_pointer_needed if -fno-omit-frame-pointer is used without
21640 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21641 with ix86_finalize_stack_frame_flags.
21642 (ix86_expand_epilogue): Likewise.
21643 (ix86_expand_split_stack_prologue): Likewise.
21644 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
21646 2017-08-10 Martin Liska <mliska@suse.cz>
21649 * c-attribs.c (handle_target_attribute):
21650 Report warning for an empty string argument of target attribute.
21652 2017-08-09 Jakub Jelinek <jakub@redhat.com>
21655 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
21657 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
21658 or DECL_NONLOCAL labels.
21659 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
21660 or DECL_NONLOCAL labels here.
21662 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
21664 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
21665 to indicate when early gimple folding has been disabled.
21666 (rs6000_gimple_fold_builtin): Add debug content.
21667 (rs6000_invalid_builtin): Fix whitespace.
21668 (rs6000_expand_builtin): Fix whitespace.
21669 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
21671 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
21674 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
21675 SAVE_MULTIPLE if not all the registers that saves, should be saved.
21677 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
21679 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
21680 (qdf24xx): Likewise.
21681 * config/aarch64/aarch64.md: Include falkor.md.
21682 * config/aarch64/falkor.md: New.
21684 2017-08-09 Marek Polacek <polacek@redhat.com>
21687 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
21688 * diagnostic.c (emit_diagnostic): Add a comment.
21689 (emit_diagnostic_valist): New function.
21691 2017-08-09 Marek Polacek <polacek@redhat.com>
21694 * input.c (make_location): New overload.
21695 * input.h (make_location): Declare.
21697 2017-08-08 Alan Modra <amodra@gmail.com>
21698 H.J. Lu <hongjiu.lu@intel.com>
21701 * gcc.c (NO_PIE_SPEC): Delete.
21702 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
21704 (LINK_PIE_SPEC): ..to here.
21705 (LINK_COMMAND_SPEC): Support -no-pie.
21706 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
21707 chain of crtbegin*.o selection, update for PIE_SPEC changes and
21709 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
21710 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
21711 (ENDFILE_CRTEND_SPEC): Similarly.
21713 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
21716 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
21717 (mstack-protector-guard-offset=): Ditto.
21718 * config/i386/i386.c (ix86_option_override): Handle
21719 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
21721 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
21722 ix86_stack_protect_guard_offset variables.
21723 (TARGET_STACK_PROTECT_GUARD): Always define.
21724 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
21725 and -mstack-protector-guard-offset= options.
21727 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21729 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
21730 boundary case for the last candidate.
21732 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21734 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
21735 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
21737 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21739 PR middle-end/19706
21740 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
21741 * config/aarch64/aarch64-builtins.c
21742 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
21743 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
21744 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
21746 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21747 Andrew Pinski <pinskia@gmail.com>
21749 PR middle-end/19706
21750 * internal-fn.def (XORSIGN): New.
21751 * optabs.def (xorsign_optab): New.
21752 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
21753 (convert_expand_mult_copysign): New.
21754 (pass_optimize_widening_mul::execute): Call
21755 convert_expand_mult_copysign.
21757 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21759 PR tree-optimization/81354
21760 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
21761 Insert on edges rather than explicitly creating landing pads.
21762 (analyze_candidates_and_replace): Commit edge inserts.
21764 2017-08-08 Richard Biener <rguenther@suse.de>
21766 PR middle-end/81719
21767 * tree-ssa-loop-niter.c: Include tree-dfa.h.
21768 (expand_simple_operations): Also look through ADDR_EXPRs with
21769 MEM_REF bases treating them as POINTER_PLUS_EXPR.
21771 2017-08-08 Richard Biener <rguenther@suse.de>
21773 PR tree-optimization/81723
21774 * tree-vect-slp.c (struct bst_traits): New hash traits.
21775 (bst_fail): New global.
21776 (vect_build_slp_tree_2): New worker, split out from ...
21777 (vect_build_slp_tree): ... this now wrapping it with using
21778 bst_fail set to cache SLP tree build fails. Properly handle
21780 (vect_analyze_slp_instance): Allocate and free bst_fail.
21782 2017-08-08 Martin Liska <mliska@suse.cz>
21785 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
21786 LABEL_DECLs that can be from a different function.
21788 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21790 PR tree-optimization/81744
21791 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
21792 loop's number of iterations.
21794 2017-08-08 Martin Liska <mliska@suse.cz>
21796 * asan.c: Include header files.
21797 * attribs.c (build_decl_attribute_variant): New function moved
21799 (build_type_attribute_qual_variant): Likewise.
21800 (cmp_attrib_identifiers): Likewise.
21801 (simple_cst_list_equal): Likewise.
21802 (omp_declare_simd_clauses_equal): Likewise.
21803 (attribute_value_equal): Likewise.
21804 (comp_type_attributes): Likewise.
21805 (build_type_attribute_variant): Likewise.
21806 (lookup_ident_attribute): Likewise.
21807 (remove_attribute): Likewise.
21808 (merge_attributes): Likewise.
21809 (merge_type_attributes): Likewise.
21810 (merge_decl_attributes): Likewise.
21811 (merge_dllimport_decl_attributes): Likewise.
21812 (handle_dll_attribute): Likewise.
21813 (attribute_list_equal): Likewise.
21814 (attribute_list_contained): Likewise.
21815 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21816 (lookup_attribute_by_prefix): Likewise.
21817 * bb-reorder.c: Include header files.
21818 * builtins.c: Likewise.
21819 * calls.c: Likewise.
21820 * cfgexpand.c: Likewise.
21821 * cgraph.c: Likewise.
21822 * cgraphunit.c: Likewise.
21823 * convert.c: Likewise.
21824 * dwarf2out.c: Likewise.
21825 * final.c: Likewise.
21826 * fold-const.c: Likewise.
21827 * function.c: Likewise.
21828 * gimple-expr.c: Likewise.
21829 * gimple-fold.c: Likewise.
21830 * gimple-pretty-print.c: Likewise.
21831 * gimple.c: Likewise.
21832 * gimplify.c: Likewise.
21833 * hsa-common.c: Likewise.
21834 * hsa-gen.c: Likewise.
21835 * internal-fn.c: Likewise.
21836 * ipa-chkp.c: Likewise.
21837 * ipa-cp.c: Likewise.
21838 * ipa-devirt.c: Likewise.
21839 * ipa-fnsummary.c: Likewise.
21840 * ipa-inline.c: Likewise.
21841 * ipa-visibility.c: Likewise.
21843 * lto-cgraph.c: Likewise.
21844 * omp-expand.c: Likewise.
21845 * omp-general.c: Likewise.
21846 * omp-low.c: Likewise.
21847 * omp-offload.c: Likewise.
21848 * omp-simd-clone.c: Likewise.
21849 * opts-global.c: Likewise.
21850 * passes.c: Likewise.
21851 * predict.c: Likewise.
21852 * sancov.c: Likewise.
21853 * sanopt.c: Likewise.
21854 * symtab.c: Likewise.
21855 * toplev.c: Likewise.
21856 * trans-mem.c: Likewise.
21857 * tree-chkp.c: Likewise.
21858 * tree-eh.c: Likewise.
21859 * tree-into-ssa.c: Likewise.
21860 * tree-object-size.c: Likewise.
21861 * tree-parloops.c: Likewise.
21862 * tree-profile.c: Likewise.
21863 * tree-ssa-ccp.c: Likewise.
21864 * tree-ssa-live.c: Likewise.
21865 * tree-ssa-loop.c: Likewise.
21866 * tree-ssa-sccvn.c: Likewise.
21867 * tree-ssa-structalias.c: Likewise.
21868 * tree-ssa.c: Likewise.
21869 * tree-streamer-in.c: Likewise.
21870 * tree-vectorizer.c: Likewise.
21871 * tree-vrp.c: Likewise.
21872 * tsan.c: Likewise.
21873 * ubsan.c: Likewise.
21874 * varasm.c: Likewise.
21875 * varpool.c: Likewise.
21876 * tree.c: Remove functions moved to attribs.[ch].
21877 * tree.h: Likewise.
21878 * config/aarch64/aarch64.c: Add attrs.h header file.
21879 * config/alpha/alpha.c: Likewise.
21880 * config/arc/arc.c: Likewise.
21881 * config/arm/arm.c: Likewise.
21882 * config/avr/avr.c: Likewise.
21883 * config/bfin/bfin.c: Likewise.
21884 * config/c6x/c6x.c: Likewise.
21885 * config/cr16/cr16.c: Likewise.
21886 * config/cris/cris.c: Likewise.
21887 * config/darwin.c: Likewise.
21888 * config/epiphany/epiphany.c: Likewise.
21889 * config/fr30/fr30.c: Likewise.
21890 * config/frv/frv.c: Likewise.
21891 * config/ft32/ft32.c: Likewise.
21892 * config/h8300/h8300.c: Likewise.
21893 * config/i386/winnt.c: Likewise.
21894 * config/ia64/ia64.c: Likewise.
21895 * config/iq2000/iq2000.c: Likewise.
21896 * config/lm32/lm32.c: Likewise.
21897 * config/m32c/m32c.c: Likewise.
21898 * config/m32r/m32r.c: Likewise.
21899 * config/m68k/m68k.c: Likewise.
21900 * config/mcore/mcore.c: Likewise.
21901 * config/microblaze/microblaze.c: Likewise.
21902 * config/mips/mips.c: Likewise.
21903 * config/mmix/mmix.c: Likewise.
21904 * config/mn10300/mn10300.c: Likewise.
21905 * config/moxie/moxie.c: Likewise.
21906 * config/msp430/msp430.c: Likewise.
21907 * config/nds32/nds32-isr.c: Likewise.
21908 * config/nds32/nds32.c: Likewise.
21909 * config/nios2/nios2.c: Likewise.
21910 * config/nvptx/nvptx.c: Likewise.
21911 * config/pa/pa.c: Likewise.
21912 * config/pdp11/pdp11.c: Likewise.
21913 * config/powerpcspe/powerpcspe.c: Likewise.
21914 * config/riscv/riscv.c: Likewise.
21915 * config/rl78/rl78.c: Likewise.
21916 * config/rx/rx.c: Likewise.
21917 * config/s390/s390.c: Likewise.
21918 * config/sh/sh.c: Likewise.
21919 * config/sol2.c: Likewise.
21920 * config/sparc/sparc.c: Likewise.
21921 * config/spu/spu.c: Likewise.
21922 * config/stormy16/stormy16.c: Likewise.
21923 * config/tilegx/tilegx.c: Likewise.
21924 * config/tilepro/tilepro.c: Likewise.
21925 * config/v850/v850.c: Likewise.
21926 * config/vax/vax.c: Likewise.
21927 * config/visium/visium.c: Likewise.
21928 * config/xtensa/xtensa.c: Likewise.
21930 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21933 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
21934 constraints since the -mupper-regs-* switches have been
21936 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
21937 into a vector from a double word element that was extracted from
21938 another vector, and eliminate extra XXPERMDI instructions.
21939 (vsx_concat_<mode>_2): Likewise.
21940 (vsx_concat_<mode>_3): Likewise.
21941 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
21942 concat to allow optimizing inserts from previous extracts.
21944 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
21946 * config/i386/i386.c (ix86_stack_protect_guard): Generate
21947 memory reference to a SSP offset in TLS address space.
21948 (ix86_print_operand) <case '@'>: Remove.
21949 (ix86_print_operand_punct_valid_p): Remove '@' code.
21950 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
21951 UNSPEC_SP_TLS_TEST.
21952 (stack_tls_protect_set_<mode>): Remove.
21953 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
21954 (stack_tls_protect_test_<mode>): Remove.
21955 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
21957 2017-08-07 Olivier Hainque <hainque@adacore.com>
21960 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
21962 2017-08-07 Douglas Rupp <rupp@adacore.com>
21964 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
21965 variable was referenced as multidir in command.
21967 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21970 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
21973 2017-08-07 Martin Liska <mliska@suse.cz>
21975 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
21976 * config/rl78/rl78.c: Add include of attribs.h.
21977 * config/sh/sh.c: Likewise.
21978 * config/v850/v850.c: Likewise.
21980 2017-08-07 Tom de Vries <tom@codesourcery.com>
21982 PR middle-end/78266
21983 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
21985 2017-08-07 Martin Liska <mliska@suse.cz>
21987 * config/mips/mips.c: Include attribs.h.
21989 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
21992 * doc/invoke.texi: Document change in behvaior for -Ofast for
21995 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
21997 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
21999 (aarch64_pop_regs): Likewise.
22000 (aarch64_gen_load_pair): Likewise.
22001 (aarch64_save_callee_saves): Likewise.
22002 (aarch64_restore_callee_saves): Likewise.
22004 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22006 * config/i386/i386.c: Revert the last change.
22008 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22011 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
22013 (ix86_finalize_stack_frame_flags): This. Also clear
22014 frame_pointer_needed if -fno-omit-frame-pointer is used without
22016 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
22017 with ix86_finalize_stack_frame_flags.
22018 (ix86_expand_epilogue): Likewise.
22019 (ix86_expand_split_stack_prologue): Likewise.
22021 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22024 * config/i386/i386.c (get_builtin_code_for_version): Set priority
22025 to P_AES for Westmere.
22027 2017-08-07 Jonathan Yong <10walls@gmail.com>
22029 * config/i386/mingw.opt (fset-stack-executable): Removed.
22030 * config/i386/cygming.opt (fset-stack-executable): Moved
22032 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
22034 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
22036 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
22038 2017-08-07 Marek Polacek <polacek@redhat.com>
22040 PR middle-end/81737
22041 * fold-const.c (fold_indirect_ref_1): Check type_domain.
22043 2017-08-07 Martin Liska <mliska@suse.cz>
22045 * attribs.h (canonicalize_attr_name): New function.
22046 (cmp_attribs): Move from c-format.c and adjusted.
22047 (is_attribute_p): Moved from tree.h.
22048 * tree-inline.c: Add new includes.
22049 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
22050 (private_is_attribute_p): Remove.
22051 (private_lookup_attribute): Likewise.
22052 (private_lookup_attribute_by_prefix): Simplify.
22053 (remove_attribute): Use is_attribute_p.
22054 * tree.h: Remove removed declarations.
22056 2017-08-07 Jakub Jelinek <jakub@redhat.com>
22058 PR middle-end/81698
22059 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
22060 instead of computing it in the function. Formatting fix.
22061 (expand_case): Don't rely on default_edge being the first edge,
22062 clear it if removing it, pass default_edge to
22063 emit_case_dispatch_table.
22064 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
22067 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
22069 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
22070 insn in the function, emit NOP after the insn.
22072 2017-08-06 Tom de Vries <tom@codesourcery.com>
22074 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
22077 2017-08-06 Tom de Vries <tom@codesourcery.com>
22079 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
22082 2017-08-04 Yury Gribov <tetra2005@gmail.com>
22084 PR tree-optimization/57371
22085 * match.pd: New pattern.
22087 2017-08-04 Marek Polacek <polacek@redhat.com>
22089 PR middle-end/81695
22090 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
22091 perform the computation in offset_int.
22093 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22095 PR tree-optimization/81136
22096 * tree-vectorizer.h: Include tree-hash-traits.h.
22097 (vec_base_alignments): New typedef.
22098 (vec_info): Add a base_alignments field.
22099 (vect_record_base_alignments): Declare.
22100 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
22102 (DR_IS_CONDITIONAL_IN_STMT): New macro.
22103 (create_data_ref): Add an is_conditional_in_stmt argument.
22104 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
22105 the is_conditional_in_stmt field.
22106 (data_ref_loc): Add an is_conditional_in_stmt field.
22107 (get_references_in_stmt): Set the is_conditional_in_stmt field.
22108 (find_data_references_in_stmt): Update call to create_data_ref.
22109 (graphite_find_data_references_in_stmt): Likewise.
22110 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
22111 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
22112 (vect_record_base_alignment): New function.
22113 (vect_record_base_alignments): Likewise.
22114 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
22115 for nested statements even if we fail to compute a misalignment.
22116 Use pooled base alignments for unconditional references.
22117 (vect_find_same_alignment_drs): Compare base addresses instead
22119 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
22120 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
22122 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22124 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
22125 Add an explicit name for the enum. Use auto_vec for slp_instances
22126 and grouped_stores.
22127 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
22129 (_bb_vec_info): Add a constructor and destructor.
22130 (vinfo_for_stmt): Return NULL for uids of -1 as well.
22131 (destroy_loop_vec_info): Delete.
22132 (vect_destroy_datarefs): Likewise.
22133 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
22134 (vec_info::vec_info): New function.
22135 (vec_info::~vec_info): Likewise.
22136 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
22137 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
22138 destroy_loop_vec_info.
22139 * tree-vect-loop.c (new_loop_vec_info): Replace with...
22140 (_loop_vec_info::_loop_vec_info): ...this.
22141 (destroy_loop_vec_info): Replace with...
22142 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
22143 the stmt_vec_infos. Leave handling of vec_info information to its
22144 destructor. Remove explicit vector releases.
22145 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
22146 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
22147 * tree-vect-slp.c (new_bb_vec_info): Replace with...
22148 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
22149 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
22150 (destroy_bb_vec_info): Replace with...
22151 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
22152 information to its destructor.
22153 (vect_slp_analyze_bb_1): Use new and delete instead of
22154 new_bb_vec_info and destroy_bb_vec_info.
22155 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
22158 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22160 * tree-data-ref.h (subscript): Add access_fn field.
22161 (data_dependence_relation): Add could_be_independent_p.
22162 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
22163 (same_access_functions): Move to tree-data-ref.c.
22164 * tree-data-ref.c (ref_contains_union_access_p): New function.
22165 (access_fn_component_p): Likewise.
22166 (access_fn_components_comparable_p): Likewise.
22167 (dr_analyze_indices): Add a reference to access_fn_component_p.
22168 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
22170 (constant_access_functions): Likewise.
22171 (add_other_self_distances): Likewise.
22172 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
22173 (initialize_data_dependence_relation): Use XCNEW and remove
22174 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
22175 of access functions that have the same type. Allow the
22176 subsequence to end with different bases in some circumstances.
22177 Record the chosen access functions in SUB_ACCESS_FN.
22178 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
22179 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
22180 (subscript_dependence_tester_1): Likewise dra and drb.
22181 (build_classic_dist_vector): Update calls accordingly.
22182 (subscript_dependence_tester): Likewise.
22183 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
22184 DDR_COULD_BE_INDEPENDENT_P.
22185 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
22186 comp_alias_ddrs instead of may_alias_ddrs.
22187 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
22189 (vect_analyze_data_ref_dependence): Use it if
22190 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
22191 distance vectors if that fails.
22192 (dependence_distance_ge_vf): New function.
22193 (vect_prune_runtime_alias_test_list): Use it. Don't clear
22194 LOOP_VINFO_MAY_ALIAS_DDRS.
22196 2017-08-04 Richard Biener <rguenther@suse.de>
22198 PR middle-end/81705
22199 * fold-const.c (fold_binary_loc): Properly restrict
22200 minus_var0 && minus_var1 case when associating undefined overflow
22203 2017-08-04 Tom de Vries <tom@codesourcery.com>
22205 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
22207 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22209 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22210 Don't start diagnostic messages with a capital letter.
22211 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22213 (rs6000_invalid_builtin): Likewise.
22214 (rs6000_trampoline_init): Likewise.
22216 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22219 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
22220 after setting changeable df flags.
22222 2017-08-03 Richard Biener <rguenther@suse.de>
22224 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
22225 up if the use is in USE - X.
22227 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
22229 * toplev.c (dumpfile.h): New include.
22230 (internal_error_reentered): New static function. Use it...
22231 (internal_error_function): ...here to handle reentered internal_error.
22233 2017-08-03 Richard Biener <rguenther@suse.de>
22235 PR middle-end/81148
22236 * fold-const.c (split_tree): Add minus_var and minus_con
22237 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
22238 here but always use minus_*.
22239 (associate_trees): Assert we never associate with MINUS_EXPR
22240 and NULL first operand. Do not recurse for PLUS_EXPR operands
22241 when associating as MINUS_EXPR either.
22242 (fold_binary_loc): Track minus_var and minus_con.
22244 2017-08-03 Tom de Vries <tom@codesourcery.com>
22247 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
22248 ACCEL_COMPILER, apply finish_options on
22249 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
22251 2017-08-03 Tom de Vries <tom@codesourcery.com>
22254 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
22255 function_entry_patch_area_size > 0.
22257 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22260 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
22261 instead of 10??LU, perform unit multiplication in wide_int,
22262 don't change alloc_object_size_limit if the limit is larger
22265 PR tree-optimization/81655
22266 PR tree-optimization/81588
22267 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
22268 the case when ranges[i].low and high are 1 for unsigned type with
22271 PR middle-end/81052
22272 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
22273 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
22275 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22277 * tree-vrp.h: Add include guard.
22279 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
22282 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
22283 (ud2): New insn pattern.
22284 * config/i386/i386.c (ix86_expand_epilogue):
22285 For naked functions, generate ud2 instead of trap insn.
22287 2017-08-02 Marek Polacek <polacek@redhat.com>
22290 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
22292 2017-08-02 Tom de Vries <tom@codesourcery.com>
22293 Cesar Philippidis <cesar@codesourcery.com>
22295 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
22296 Add missing edge probabilities.
22298 2017-08-02 Tamar Christina <tamar.christina@arm.com>
22300 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
22301 Correct endianness.
22303 2017-08-02 Jakub Jelinek <jakub@redhat.com>
22305 PR middle-end/79499
22306 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
22307 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
22308 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
22309 prologue_seq sequences - if any.
22311 2017-08-02 Richard Biener <rguenther@suse.de>
22313 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
22314 via vectors if supported, integer extracts via punning if supported
22315 or otherwise vector extracts.
22317 2017-08-02 Richard Biener <rguenther@suse.de>
22319 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
22321 (bitmap_insert_into_set): ... this.
22323 2017-08-02 Richard Biener <rguenther@suse.de>
22325 PR tree-optimization/81633
22327 2015-08-17 Alan Hayward <alan.hayward@arm.com>
22329 PR tree-optimization/71752
22330 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
22332 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
22334 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
22335 (machine_function::call_ms2sysv_pad_out): Remove field.
22336 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
22337 (ix86_compute_frame_layout): Likewise.
22339 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
22342 * config/i386/i386.c (ix86_set_func_type): Disallow naked
22343 attribute with interrupt attribute.
22345 2017-08-01 Andrew Pinski <apinski@cavium.com>
22347 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
22348 BIT_INSERT_EXPR's operand 1
22349 to see if the types precision matches.
22351 2017-08-01 Martin Liska <mliska@suse.cz>
22353 PR middle-end/70140
22354 * builtins.c (expand_builtin_memcpy_args): Remove.
22355 (expand_builtin_memcpy): Call newly added function
22356 expand_builtin_memory_copy_args.
22357 (expand_builtin_memcpy_with_bounds): Likewise.
22358 (expand_builtin_mempcpy): Remove last argument.
22359 (expand_builtin_mempcpy_with_bounds): Likewise.
22360 (expand_builtin_memory_copy_args): New function created from
22361 expand_builtin_mempcpy_args with small modifications.
22362 (expand_builtin_mempcpy_args): Remove.
22363 (expand_builtin_stpcpy): Remove unused argument.
22364 (expand_builtin): Likewise.
22365 (expand_builtin_with_bounds): Likewise.
22367 2017-08-01 Martin Liska <mliska@suse.cz>
22370 Make mempcpy more optimal (PR middle-end/70140).
22372 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22375 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
22376 __builtin_vec_cmpne verify both arguments are compatible vectors
22377 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
22378 verify arg1_type is a pointer or array type. For __builtin_vec_st,
22379 move computation of aligned to after checking the argument types.
22383 * config/rs6000/vsx.md (vextract_fp_from_shorth,
22384 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
22387 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
22388 Doug Rupp <rupp@adacore.com>
22389 Olivier Hainque <hainque@adacore.com>
22391 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
22392 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
22394 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
22395 for TARGET_OS_CPP_BUILTIN.
22396 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
22397 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
22399 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
22400 passing required abi options to the assembler for EABI configurations.
22401 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
22402 of .text.hot and .text.unlikely sections for kernel modules when
22403 using ARM style exceptions.
22404 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
22405 options. Add EXTRA_CC1_SPEC.
22406 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
22408 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
22410 (ARM_TARGET2_DWARF_FORMAT): Define.
22411 * config/arm/t-vxworks: Adjust multilib control to removal of the
22412 Diab command line options.
22414 2017-08-01 Martin Liska <mliska@suse.cz>
22416 PR gcov-profile/81561
22417 * gcov.c (unblock): Make unblocking safe as we need to preserve
22418 index correspondence of blocks and block_lists.
22420 2017-08-01 Richard Biener <rguenther@suse.de>
22422 PR tree-optimization/81181
22423 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
22424 (compute_antic): ... end of iteration here.
22426 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
22428 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
22429 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
22430 (ftree-slp-vectorize): Likewise.
22431 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
22432 can no longer be set independent of flag_tree_loop_vectorize.
22433 * omp-general.c (emp_max_vf): Likewise.
22434 * opts.c (enable_fdo_optimizations): Remove references to
22435 flag_tree_vectorize, these are now implicit.
22436 (common_handle_option): Remove handling for OPT_ftree_vectorize,
22437 and leave it for the options machinery.
22439 2017-08-01 Martin Liska <mliska@suse.cz>
22441 PR middle-end/70140
22442 * builtins.c (expand_builtin_memcpy_args): Remove.
22443 (expand_builtin_memcpy): Call newly added function
22444 expand_builtin_memory_copy_args.
22445 (expand_builtin_memcpy_with_bounds): Likewise.
22446 (expand_builtin_mempcpy): Remove last argument.
22447 (expand_builtin_mempcpy_with_bounds): Likewise.
22448 (expand_builtin_memory_copy_args): New function created from
22449 expand_builtin_mempcpy_args with small modifications.
22450 (expand_builtin_mempcpy_args): Remove.
22451 (expand_builtin_stpcpy): Remove unused argument.
22452 (expand_builtin): Likewise.
22453 (expand_builtin_with_bounds): Likewise.
22455 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22458 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
22459 print "ds:" only for immediates in generic address space.
22461 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22464 * config/i386/i386.c (ix86_funciton_naked): New prototype.
22465 (ix86_function_ok_for_sibcall): Return false for naked functions.
22467 2017-08-01 Richard Biener <rguenther@suse.de>
22469 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
22470 (compute_antic): Seed worklist with exit block predecessors.
22471 * cfganal.c (dfs_find_deadend): For a cycle return the source
22472 of the edge closing it.
22474 2017-08-01 Tamar Christina <tamar.christina@arm.com>
22476 * config/aarch64/aarch64.c
22477 (aarch64_can_const_movi_rtx_p): Move 0 check.
22479 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22481 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
22482 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
22484 * match.pd: Ditto in address comparison pattern.
22486 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22488 PR tree-optimization/81627
22489 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
22490 closed ssa form for store-store chain.
22492 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22494 PR tree-optimization/81620
22495 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
22496 for store-store chain.
22498 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22500 PR tree-optimization/81588
22501 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
22502 ranges[i].in_p, invert comparison code ccode. For >/>=,
22503 swap rhs1 and rhs2 and comparison code unconditionally,
22504 for </<= don't do that. Don't swap rhs1/rhs2 again if
22505 ranges[i].in_p, instead invert comparison code ccode if
22506 opcode or oe->rank is BIT_IOR_EXPR.
22509 * optabs.def (vec_extract_optab, vec_init_optab): Change from
22510 a direct optab to conversion optab.
22511 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
22512 with GET_MODE_INNER as last argument instead of optab_handler.
22513 * expmed.c (extract_bit_field_1): Likewise. Use vector from
22514 vector extraction if possible and optab is available.
22515 * expr.c (store_constructor): Use convert_optab_handler instead
22516 of optab_handler. Use vector initialization from smaller
22517 vectors if possible and optab is available.
22518 * tree-vect-stmts.c (vectorizable_load): Likewise.
22519 * doc/md.texi (vec_extract, vec_init): Document that the optabs
22520 now have two modes.
22521 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
22522 of vec_init from half-sized vectors with the same element mode.
22523 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
22524 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
22525 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
22526 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
22527 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
22528 after mode in gen_vec_extract* calls.
22529 (vec_extract<mode>): Renamed to ...
22530 (vec_extract<mode><ssescalarmodelower>): ... this.
22531 (vec_extract<mode><ssehalfvecmodelower>): New expander.
22532 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
22533 element mode after mode in gen_vec_init* calls.
22534 (VEC_INIT_HALF_MODE): New mode iterator.
22535 (vec_init<mode>): Renamed to ...
22536 (vec_init<mode><ssescalarmodelower>): ... this.
22537 (vec_init<mode><ssehalfvecmodelower>): New expander.
22538 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
22539 (vec_extractv2sfsf): ... this.
22540 (vec_initv2sf): Renamed to ...
22541 (vec_initv2sfsf): ... this.
22542 (vec_extractv2si): Renamed to ...
22543 (vec_extractv2sisi): ... this.
22544 (vec_initv2si): Renamed to ...
22545 (vec_initv2sisi): ... this.
22546 (vec_extractv4hi): Renamed to ...
22547 (vec_extractv4hihi): ... this.
22548 (vec_initv4hi): Renamed to ...
22549 (vec_initv4hihi): ... this.
22550 (vec_extractv8qi): Renamed to ...
22551 (vec_extractv8qiqi): ... this.
22552 (vec_initv8qi): Renamed to ...
22553 (vec_initv8qiqi): ... this.
22554 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
22555 (vec_init<mode>): Renamed to ...
22556 (vec_init<mode><VEC_base_l>): ... this.
22557 (vec_extract<mode>): Renamed to ...
22558 (vec_extract<mode><VEC_base_l>): ... this.
22559 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
22560 (vec_initv2sfsf): ... this.
22561 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
22562 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22563 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
22564 element mode after mode in gen_vec_init* calls.
22565 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
22566 (vec_init<mode><Vel>): ... this.
22567 (vec_extract<mode>): Renamed to ...
22568 (vec_extract<mode><Vel>): ... this.
22569 * config/aarch64/iterators.md (Vel): New mode attribute.
22570 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
22571 Add element mode after mode in gen_vec_extract* calls.
22572 * config/s390/vector.md (non_vec_l): New mode attribute.
22573 (vec_extract<mode>): Renamed to ...
22574 (vec_extract<mode><non_vec_l>): ... this.
22575 (vec_init<mode>): Renamed to ...
22576 (vec_init<mode><non_vec_l>): ... this.
22577 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
22578 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
22580 * config/arm/iterators.md (V_elem_l): New mode attribute.
22581 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
22582 (vec_extract<mode><V_elem_l>): ... this.
22583 (vec_extractv2di): Renamed to ...
22584 (vec_extractv2didi): ... this.
22585 (vec_init<mode>): Renamed to ...
22586 (vec_init<mode><V_elem_l>): ... this.
22587 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
22588 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
22589 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
22590 Add element mode after gen_vec_extract* calls.
22591 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
22592 (vec_init<mode><unitmode>): ... this.
22593 (vec_extract<mode>): Renamed to ...
22594 (vec_extract<mode><unitmode>): ... this.
22595 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
22596 (vec_init<mode><unitmode>): ... this.
22597 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
22598 (vec_initv2sfsf): ... this.
22599 (vec_extractv2sf): Renamed to ...
22600 (vec_extractv2sfsf): ... this.
22601 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
22602 Add element mode after gen_vec_extract* calls.
22603 * config/mips/mips.md (unitmode): New mode iterator.
22604 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
22605 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
22606 * config/spu/spu.md (inner_l): New mode attribute.
22607 (vec_init<mode>): Renamed to ...
22608 (vec_init<mode><inner_l>): ... this.
22609 (vec_extract<mode>): Renamed to ...
22610 (vec_extract<mode><inner_l>): ... this.
22611 * config/sparc/sparc.md (veltmode): New mode iterator.
22612 (vec_init<VMALL:mode>): Renamed to ...
22613 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
22614 * config/ia64/vect.md (vec_initv2si): Renamed to ...
22615 (vec_initv2sisi): ... this.
22616 (vec_initv2sf): Renamed to ...
22617 (vec_initv2sfsf): ... this.
22618 (vec_extractv2sf): Renamed to ...
22619 (vec_extractv2sfsf): ... this.
22620 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
22621 (vec_init<mode>): Renamed to ...
22622 (vec_init<mode><VEC_base_l>): ... this.
22623 (vec_extract<mode>): Renamed to ...
22624 (vec_extract<mode><VEC_base_l>): ... this.
22625 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
22626 (vec_initv2sfsf): ... this.
22627 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
22628 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22629 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
22630 gen_vec_init* calls.
22632 2017-08-01 Richard Biener <rguenther@suse.de>
22634 PR tree-optimization/81297
22635 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
22636 TREE_OVERFLOW from INTEGER_CSTs.
22638 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
22640 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
22642 2017-07-31 Carl Love <cel@us.ibm.com>
22644 * config/rs6000/rs6000-c: Add support for built-in functions
22645 vector signed char vec_xl_be (signed long long, signed char *);
22646 vector unsigned char vec_xl_be (signed long long, unsigned char *);
22647 vector signed int vec_xl_be (signed long long, signed int *);
22648 vector unsigned int vec_xl_be (signed long long, unsigned int *);
22649 vector signed long long vec_xl_be (signed long long, signed long long *);
22650 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
22651 vector signed short vec_xl_be (signed long long, signed short *);
22652 vector unsigned short vec_xl_be (signed long long, unsigned short *);
22653 vector double vec_xl_be (signed long long, double *);
22654 vector float vec_xl_be (signed long long, float *);
22655 * config/rs6000/altivec.h (vec_xl_be): Add #define.
22656 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
22657 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
22659 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
22660 (altivec_expand_builtin): Add switch statement to call
22661 altivec_expand_xl_be for each builtin.
22662 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
22663 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
22664 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
22665 __builtin_vsx_le_be_v16qi.
22666 * doc/extend.texi: Update the built-in documentation file for the
22667 new built-in functions.
22669 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22672 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
22674 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22676 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22678 * config.gcc: Add z14.
22679 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
22680 CPU model numbers for z13s and z14.
22681 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
22683 * config/s390/s390-opts.h (enum processor_type): Rename
22684 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22685 * config/s390/s390.c (processor_table): Add field for CPU name to
22686 be passed to Binutils.
22687 (s390_asm_output_machine_for_arch): Use the new field in
22688 processor_table for Binutils.
22689 (s390_expand_builtin): Replace arch12 with z14.
22690 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22691 (s390_get_sched_attrmask): Likewise.
22692 (s390_get_unit_mask): Likewise.
22693 * config/s390/s390.opt: Add z14 to processor_type enum.
22695 2017-07-31 Martin Jambor <mjambor@suse.cz>
22698 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
22699 regardless of optimization level.
22701 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
22702 Martin Liska <mliska@suse.cz>
22704 * predict.def: Remove old comment and adjust probability.
22705 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
22706 PREDICT statements.
22708 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22711 * config/i386/i386.c (ix86_function_naked): New function.
22712 (ix86_can_use_return_insn_p): Return false for naked functions.
22713 (ix86_expand_prologue): Skip prologue for naked functions.
22714 (ix86_expand_epilogue): Skip epilogue for naked functions
22715 and emit trap instruction.
22716 (ix86_warn_func_return): New function.
22717 (ix86_attribute_table): Add "naked" attribute specification.
22718 (TARGET_WARN_FUNC_RETURN): Define.
22719 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
22721 2017-07-31 Martin Liska <mliska@suse.cz>
22723 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
22724 (dump_gimple_bb_header): Always dump BB info.
22725 (pp_cfg_jump): Do not append info about BB when dumping a jump.
22727 2017-07-31 Martin Liska <mliska@suse.cz>
22730 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
22731 also with current_function_decl non-null equality.
22733 2017-07-31 Jakub Jelinek <jakub@redhat.com>
22736 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
22737 change type to the element type, instead add eltype variable and
22738 use it where we are interested in the element type.
22740 PR tree-optimization/81603
22741 * ipa-polymorphic-call.c
22742 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
22743 offset arithmetic in offset_int, bail out if the resulting bit offset
22744 doesn't fit into shwi.
22746 2017-07-31 Martin Liska <mliska@suse.cz>
22748 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
22749 (gimplify_save_expr): Fix comment.
22751 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
22754 * config/i386/i386.c (ix86_function_arg): Update arguments for
22756 (ix86_compute_frame_layout): Set the initial stack offset to
22757 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
22758 INCOMING_FRAME_SP_OFFSET.
22759 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
22760 stack before exception handler returns.
22761 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
22762 the 'ERROR_CODE' for exception handler.
22764 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
22766 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
22767 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
22768 (ASM_OUTPUT_REG_POP): Ditto.
22769 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
22770 instead of asm_fprintf to output pure string.
22772 2017-07-29 Jakub Jelinek <jakub@redhat.com>
22774 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
22775 to imported_module_or_decl hook.
22776 (debug_nothing_tree_tree_tree_bool): Remove.
22777 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
22778 * debug.c (do_nothing_debug_hooks): Use
22779 debug_nothing_tree_tree_tree_bool_bool instead of
22780 debug_nothing_tree_tree_tree_bool.
22781 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
22782 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
22783 * sdbout.c (sdb_debug_hooks): Likewise.
22784 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
22785 (gen_namespace_die): Add DW_AT_export_symbols attribute if
22787 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
22788 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
22789 attribute, don't add anything.
22791 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22793 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
22794 (fold_build2_stat_loc): Likewise.
22795 (fold_build3_stat_loc): Likewise.
22796 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
22797 (fold_build1_loc): Remove macro.
22798 (fold_build2_loc): Likewise.
22799 (fold_build3_loc): Likewise.
22801 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22803 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
22804 (gimple_build_debug_bind_source_stat): Likewise.
22805 * gimple.h (gimple_build_debug_bind): Remove macro.
22806 (gimple_build_debug_bind_source): Likewise.
22808 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22810 * bitmap.c (bitmap_alloc): Adjust.
22811 (bitmap_gc_alloc): Likewise.
22812 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22814 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22816 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22817 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22818 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22819 (bitmap_gc_alloc_stat): Likewise.
22820 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22822 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22824 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22825 * rtl.h (shallow_copy_rtx): Remove macro.
22827 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22829 * emit-rtl.c (gen_raw_REG): Adjust.
22830 * gengenrtl.c (gendef): Likewise.
22831 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22832 * rtl.h (rtx_alloc): Remove macro.
22834 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22836 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22837 (build_tree_list_stat): Likewise.
22838 * tree.h (build_tree_list): Remove macro.
22839 (build_tree_list_vec): Likewise.
22841 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22843 * tree.c (make_vector_stat): Remove _stat from name.
22844 (build_vector_stat): Likewise.
22845 * tree.h (make_vector_stat): Remove macro.
22846 (build_vector_stat): Likewise.
22848 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22850 * tree.h (build_var_debug_value): Remove prototype.
22852 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22854 * tree.c (tree_cons_stat): Remove _stat from name.
22855 * tree.h (tree_cons): Remove macro.
22857 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22859 * tree.c (build_vl_exp_stat): Remove _stat from name.
22860 * tree.h (build_vl_exp): Remove macro.
22862 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22864 * tree.c (build_decl_stat): Remove _stat from name.
22865 * tree.h (build_decl): Remove macro.
22867 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22869 * gimple.c (gimple_build_with_ops_stat): Adjust.
22870 (gimple_alloc_stat): Remove _stat from name.
22871 * gimple.h (gimple_alloc): Remove macro.
22873 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22875 * tree.c (make_tree_vec_stat): Remove _stat from name.
22876 (grow_tree_vec_stat): Likewise.
22877 * tree.h (make_tree_vec_stat): Adjust prototype.
22878 (grow_tree_vec_stat): Likewise.
22879 (make_tree_vec): Remove macro.
22880 (grow_tree_vec): Likewise.
22882 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22884 * fold-const.c (fold_build1_stat_loc): Adjust.
22885 (fold_build2_stat_loc): Likewise.
22886 (fold_build3_stat_loc): Likewise.
22887 * tree.c (build0_stat): Remove _stat from name.
22888 (build1_stat): Likewise.
22889 (build2_stat): Likewise.
22890 (build3_stat): Likewise.
22891 (build4_stat): Likewise.
22892 (build5_stat): Likewise.
22893 * tree.h (build1_loc): Remove macro, and rename _stat function
22895 (build2_loc): Likewise.
22896 (build3_loc): Likewise.
22897 (build4_loc): Likewise.
22898 (build5_loc): Likewise.
22900 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22902 * tree.c (make_int_cst_stat): Remove _stat from name.
22903 * tree.h (make_int_cst_stat): Adjust prototype.
22904 (make_int_cst): Remove macro.
22906 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22908 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22909 * tree.h (make_tree_binfo_stat): Adjust prototype.
22910 (make_tree_binfo): Remove.
22912 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22914 * tree.c (copy_node_stat): Rename to copy_node.
22915 (build_distinct_type_copy): Adjust.
22916 * tree.h (copy_node_stat): Adjust prototype.
22917 (copy_node): Remove macro.
22919 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22921 * tree.c (make_node_stat): rename to make_node.
22922 (build_tree_list_stat): Adjust.
22923 (build0_stat): Likewise.
22924 (build2_stat): Likewise.
22925 (build3_stat): Likewise.
22926 (build4_stat): Likewise.
22927 (build5_stat): Likewise.
22928 (build_decl_stat): Likewise.
22929 * tree.h (make_node_stat): Adjust prototype.
22930 (make_node): remove macro.
22932 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
22934 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
22935 (PPC_FEATURE2_SCV): Likewise.
22936 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
22938 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22940 * config/aarch64/aarch64.c
22941 (aarch64_internal_mov_immediate): Add new special pattern.
22942 * config/aarch64/aarch64.md (*movdi_aarch64):
22943 Add reg/32bit const mov case.
22945 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22946 Richard Sandiford <richard.sandiford@linaro.org>
22948 * config/aarch64/aarch64.md (mov<mode>): Generalize.
22949 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
22950 Add integer and movi cases.
22951 (movi-split-hf-df-sf split, fp16): New.
22952 (enabled): Added TARGET_FP_F16INST.
22953 * config/aarch64/iterators.md (GPF_HF): New.
22954 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
22956 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22958 * config/aarch64/aarch64.c
22959 (aarch64_simd_container_mode): Add prototype.
22960 (aarch64_expand_mov_immediate): Add HI support.
22961 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
22962 (aarch64_can_const_movi_rtx_p): New.
22963 (aarch64_preferred_reload_class):
22964 Remove restrictions of using FP registers for certain SIMD operations.
22965 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
22966 (aarch64_valid_floating_const): Add integer move validation.
22967 (aarch64_simd_imm_scalar_p): Remove.
22968 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
22969 (aarch64_legitimate_constant_p): Expand list of supported cases.
22970 * config/aarch64/aarch64-protos.h
22971 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
22972 (aarch64_reinterpret_float_as_int): New.
22973 (aarch64_simd_imm_scalar_p): Remove.
22974 * config/aarch64/constraints.md (Uvi): New.
22975 (Dd): Split into Ds and new Dd.
22976 * config/aarch64/aarch64.md (*movsi_aarch64):
22978 (*movdi_aarch64): Add SIMD mov case.
22980 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22982 * tree-predcom.c: (struct chain): Handle store-store chain in which
22983 stores for elimination only store loop invariant values.
22984 (execute_pred_commoning_chain): Ditto.
22985 (prepare_initializers_chain_store_elim): Ditto.
22986 (prepare_finalizers): Ditto.
22987 (is_inv_store_elimination_chain): New function.
22988 (initialize_root_vars_store_elim_1): New function.
22990 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22992 * tree-predcom.c: Revise general description of the pass.
22993 (enum chain_type): New enum type for store elimination.
22994 (struct chain): New field supporting store elimination.
22995 (struct component): Ditto.
22996 (dump_chain): Dump store-stores chain.
22997 (release_chain): Release resources.
22998 (split_data_refs_to_components): Compute and create component
22999 contains only stores for elimination.
23000 (get_chain_last_ref_at): New function.
23001 (make_invariant_chain): Initialization.
23002 (make_rooted_chain): Specify chain type in parameter and record it.
23003 (add_looparound_copies): Skip for store-stores chain.
23004 (determine_roots_comp): Compute type of chain and pass it to
23006 (initialize_root_vars_store_elim_2): New function.
23007 (finalize_eliminated_stores): New function.
23008 (remove_stmt): Handle store for elimination.
23009 (execute_pred_commoning_chain): Execute predictive commoning on
23010 store-store chains.
23011 (determine_unroll_factor): Skip unroll for store-stores chain.
23012 (prepare_initializers_chain_store_elim): New function.
23013 (prepare_initializers_chain): Hanlde store-store chain.
23014 (prepare_finalizers_chain, prepare_finalizers): New function.
23015 (tree_predictive_commoning_loop): Return integer value indicating
23016 if loop is unrolled or lcssa form is corrupted.
23017 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
23019 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23021 * tree-predcom.c (initialize_root): Delete.
23022 (execute_pred_commoning_chain): Initialize root vars and replace
23023 reference of non-combined chain directly, rather than call above
23026 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23028 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
23029 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
23031 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23033 * tree-predcom.c (struct chain): New field init_seq.
23034 (release_chain): Release init_seq.
23035 (prepare_initializers_chain): Record intialization stmts in above
23037 (insert_init_seqs): New function.
23038 (tree_predictive_commoning_loop): Call insert_init_seqs.
23040 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23042 * tree-predcom.c (determine_roots_comp): Skip trivial components.
23044 2017-07-28 Richard Biener <rguenther@suse.de>
23046 * match.pd: Remove superfluous :c.
23047 * genmatch.c (simplify::id): Add member.
23048 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
23050 (current_id): New global.
23051 (dt_node::parent): Move from ...
23052 (dt_operand::parent): ... here. Add for_id member.
23053 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
23054 (decision_tree::find_node): Relax order requirement when
23055 merging DT_TRUE nodes to ones inbetween the current simplify
23056 and the one we try to merge with. Add diagnostic whenever
23057 we need to enforce pattern order by not merging.
23058 (decision_tree::insert): Set current_id.
23059 (decision_tree::print_node): Dump parent node and for_id.
23060 (parser::last_id): Add member.
23061 (parser::push_simplify): Assign unique id.
23062 (parser::parser): Initialize last_id.
23064 2017-07-28 Martin Liska <mliska@suse.cz>
23067 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
23068 gimple_build_debug_bind.
23070 2017-07-28 Richard Biener <rguenther@suse.de>
23072 PR tree-optimization/81502
23073 * match.pd: Add pattern combining BIT_INSERT_EXPR with
23075 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
23077 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
23078 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
23079 for BIT_FIELD_REF args.
23080 * fold-const.c (make_bit_field_ref): Likewise.
23081 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
23083 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23086 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
23087 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
23088 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
23089 Or it into SANITIZER_UNDEFINED.
23090 * ubsan.c: Include gimple-fold.h and varasm.h.
23091 (ubsan_expand_ptr_ifn): New function.
23092 (instrument_pointer_overflow): New function.
23093 (maybe_instrument_pointer_overflow): New function.
23094 (instrument_object_size): Formatting fix.
23095 (pass_ubsan::execute): Call instrument_pointer_overflow
23096 and maybe_instrument_pointer_overflow.
23097 * internal-fn.c (expand_UBSAN_PTR): New function.
23098 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
23099 * sanitizer.def (__ubsan_handle_pointer_overflow,
23100 __ubsan_handle_pointer_overflow_abort): New builtins.
23101 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
23102 * internal-fn.def (UBSAN_PTR): New internal function.
23103 * opts.c (sanitizer_opts): Add pointer-overflow.
23104 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
23105 * fold-const.c (build_range_check): Compute pointer range check in
23106 integral type if pointer arithmetics would be needed. Formatting
23109 2017-07-28 Martin Liska <mliska@suse.cz>
23112 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
23113 parameters that are of a variable-length.
23115 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
23117 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
23119 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
23120 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
23121 (CRT_CALL_STATIC_FUNCTION): Likewise.
23122 (ASM_DEFAULT_SPEC): New define.
23123 (ASM_SPEC32): Likewise.
23124 (ASM_SPEC64): Likewise.
23125 (ASM_SPEC_COMMON): Likewise.
23126 (ASM_SPEC): Likewise.
23127 (INVALID_64BIT): Likewise.
23128 (LINK_OS_DEFAULT_SPEC): Likewise.
23129 (LINK_OS_SPEC32): Likewise.
23130 (LINK_OS_SPEC64): Likewise.
23131 (POWERPC_LINUX): Likewise.
23132 (PTRDIFF_TYPE): Likewise.
23133 (RESTORE_FP_PREFIX): Likewise.
23134 (RESTORE_FP_SUFFIX): Likewise.
23135 (SAVE_FP_PREFIX): Likewise.
23136 (SAVE_FP_SUFFIX): Likewise.
23137 (SIZE_TYPE): Likewise.
23138 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
23139 (TARGET_64BIT): Likewise.
23140 (TARGET_64BIT): Likewise.
23141 (TARGET_AIX): Likewise.
23142 (WCHAR_TYPE_SIZE): Likewise.
23143 (WCHAR_TYPE): Undefine.
23144 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
23145 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
23146 (CPP_OS_RTEMS_SPEC): Delete.
23147 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
23148 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
23150 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
23152 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23154 PR tree-optimization/81578
23155 * tree-parloops.c (build_new_reduction): Bail out if
23156 reduction_code isn't one of the standard OpenMP reductions.
23157 Move the details printing after that decision.
23159 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23161 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
23162 related to reload_in_progress.
23163 (splat_input_operand): Likewise.
23164 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
23166 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
23168 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
23169 (TARGET_INSTANTIATE_DECLS): Likewise.
23170 (legitimate_indexed_address_p): Delete reload_in_progress code.
23171 (rs6000_debug_legitimate_address_p): Likewise.
23172 (rs6000_eliminate_indexed_memrefs): Likewise.
23173 (rs6000_emit_le_vsx_store): Likewise.
23174 (rs6000_emit_move_si_sf_subreg): Likewise.
23175 (rs6000_emit_move): Likewise.
23176 (register_to_reg_type): Likewise.
23177 (rs6000_pre_atomic_barrier): Likewise.
23178 (rs6000_machopic_legitimize_pic_address): Likewise.
23179 (rs6000_allocate_stack_temp): Likewise.
23180 (rs6000_address_for_fpconvert): Likewise.
23181 (rs6000_address_for_altivec): Likewise.
23182 (rs6000_secondary_memory_needed_rtx): Delete function.
23183 (rs6000_check_sdmode): Likewise.
23184 (rs6000_alloc_sdmode_stack_slot): Likewise.
23185 (rs6000_instantiate_decls): Likewise.
23186 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
23187 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
23188 Delete reload_in_progress.
23189 (*vec_reload_and_plus_<mptrsize>): Likewise.
23190 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
23191 (vsx_div_v2di): Likewise.
23192 (vsx_udiv_v2di): Likewise.
23194 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23196 * config/rs6000/rs6000.opt (mlra): Replace with stub.
23197 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
23198 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
23199 (rs6000_debug_reg_global): Delete print of LRA status.
23200 (rs6000_option_override_internal): Delete dead LRA related code.
23201 (rs6000_lra_p): Delete function.
23202 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
23204 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
23206 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
23207 * config/riscv/rtems.h: New file.
23209 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23210 Sudakshina Das <sudi.das@arm.com>
23212 * config/aarch64/aarch64.md
23213 (define_split for and<mode>3nr_compare): Move
23214 non aarch64_logical_operand to a register.
23215 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
23216 register immediate operand to a register.
23217 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
23219 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23221 PR middle-end/81564
23222 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
23224 2017-07-27 Richard Biener <rguenther@suse.de>
23226 PR tree-optimization/81573
23227 PR tree-optimization/81494
23228 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
23229 multi defuse cycle case.
23231 2017-07-27 Richard Biener <rguenther@suse.de>
23233 PR tree-optimization/81571
23234 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
23237 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
23239 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
23240 earlier and only if MASK_FPU is set. Adjust formatting.
23242 2017-07-27 Martin Liska <mliska@suse.cz>
23244 * opt-functions.awk: Add validation of value of Init.
23245 * optc-gen.awk: Pass new argument.
23247 2017-07-27 Martin Liska <mliska@suse.cz>
23249 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
23250 Fix wrong condition.
23252 2017-07-27 Martin Liska <mliska@suse.cz>
23254 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
23255 BBs and edges seen by autoFDO.
23257 2017-07-27 Richard Biener <rguenther@suse.de>
23259 PR tree-optimization/81502
23260 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
23261 with incompatible but same sized type.
23262 (execute_update_addresses_taken): Likewise.
23264 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
23266 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
23267 flag_tree_loop_vectorize rather than flag_tree_vectorize.
23269 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23272 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
23273 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
23274 Change s_operand to memory_operand.
23276 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
23278 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
23279 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
23280 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
23281 Emit instructions rather than returning an expression. Handle TFmode
23282 and KFmode by casting to TImode.
23283 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
23284 (rs6000_emit_le_vsx_store): Likewise.
23285 * config/rs6000/vsx.md (VSX_TI): New iterator.
23286 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
23287 (*vsx_le_undo_permute_<mode>): Likewise.
23288 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
23289 emit the split sequence.
23290 (*vsx_le_perm_store_<mode>): Likewise.
23292 2017-07-27 Jakub Jelinek <jakub@redhat.com>
23294 PR tree-optimization/81555
23295 PR tree-optimization/81556
23296 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
23297 if true, force CHANGED for the recursive invocation.
23298 (reassociate_bb): Remember original length of ops array, pass
23299 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
23301 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
23302 attributes for noipa attribute. For naked attribute use
23303 lookup_attribute first before lookup_attribute_spec.
23304 * final.c (rest_of_handle_final): Disable IPA RA for functions with
23306 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
23307 for functions with noipa attribute.
23308 (cgraph_externally_visible_p): Return true for functions with noipa
23310 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
23311 for functions with noipa attribute.
23312 * doc/extend.texi: Document noipa function attribute.
23313 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
23314 also for functions with noipa attribute.
23315 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
23317 2017-07-26 Andrew Pinski <apinski@cavium.com>
23319 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
23320 vec_unalign_load_cost and vec_unalign_store_cost.
23322 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
23324 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
23325 -mvsx-small-integer option.
23326 (ISA_3_0_MASKS_IEEE): Likewise.
23327 (OTHER_VSX_VECTOR_MASKS): Likewise.
23328 (POWERPC_MASKS): Likewise.
23329 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
23330 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
23331 code, only testing for DImode being allowed in non-VSX floating
23333 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
23334 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
23336 (rs6000_option_override_internal): Delete -mvsx-small-integer.
23337 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
23338 TARGET_P8_VECTOR test.
23339 (rs6000_secondary_reload_simple_move): Likewise.
23340 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
23341 since TARGET_P9_VECTOR was already tested.
23342 (rs6000_opt_masks): Remove -mvsx-small-integer.
23343 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
23344 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23346 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
23347 test for TARGET_VEXTRACTUB was used, and that uses
23349 (p9 extract splitter): Likewise.
23350 (vsx_extract_<mode>_di_p9): Likewise.
23351 (vsx_extract_<mode>_store_p9): Likewise.
23352 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
23353 for TARGET_P9_VECTOR was used. Delete code that is now dead with
23354 the elimination of TARGET_VSX_SMALL_INTEGER.
23355 (vsx_extract_<mode>_p8): Likewise.
23356 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
23357 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
23358 (vsx_set_<mode>_p9): Likewise.
23359 (vsx_set_v4sf_p9): Likewise.
23360 (vsx_set_v4sf_p9_zero): Likewise.
23361 (vsx_insert_extract_v4sf_p9): Likewise.
23362 (vsx_insert_extract_v4sf_p9_2): Likewise.
23363 * config/rs6000/rs6000.md (sign extend splitter): Change
23364 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
23365 (floatsi<mode>2_lfiwax_mem): Likewise.
23366 (floatunssi<mode>2_lfiwzx_mem): Likewise.
23367 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
23368 since a test for TARGET_P9_VECTOR was used.
23369 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23370 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
23371 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23372 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23373 TARGET_P8_VECTOR test.
23374 (fix_trunc<mode>si2_stfiwx): Likewise.
23375 (fix_trunc<mode>si2_internal): Likewise.
23376 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
23377 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23379 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23380 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23381 TARGET_P8_VECTOR test.
23382 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23383 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
23384 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23386 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23387 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
23388 since a test for TARGET_P9_VECTOR was used.
23389 (splitter for loading small constants): Likewise.
23391 2017-07-26 Andrew Pinski <apinski@cavium.com>
23393 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
23396 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
23399 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
23400 (fp_valid_at): Likewise.
23402 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23404 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
23405 (qdf24xx_addrcost_table): Likewise.
23406 (cortexa57_tunings): Update to use generic_branch_cost.
23407 (cortexa72_tunings): Likewise.
23408 (cortexa73_tunings): Likewise.
23409 (qdf24xx_tunings): Likewise.
23411 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23413 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
23414 (thunderx2t99_branch_cost): Likewise.
23415 (cortexa35_tunings): Update to use generic_branch_cost.
23416 (cortexa53_tunings): Likewise.
23417 (cortexa57_tunings): Likewise.
23418 (cortexa72_tunings): Likewise.
23419 (cortexa73_tunings): Likewise.
23420 (thunderx2t99_tunings): Likewise.
23422 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23424 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
23425 (sparc_option_override): Honour MASK_FSMULD.
23426 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
23427 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
23428 * config/sparc/sparc.opt (mfsmuld): New option.
23429 * doc/invoke.texi (mfsmuld): Document option.
23431 2017-07-26 Marek Polacek <polacek@redhat.com>
23433 PR middle-end/70992
23434 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
23436 2017-07-26 Richard Biener <rguenther@suse.de>
23438 * gimple-match-head.c (do_valueize): Return OP if valueize
23440 (get_def): New helper to get at the def stmt of a SSA name
23441 if valueize allows.
23442 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
23443 do_valueize to get at the def stmt.
23444 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
23446 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
23448 PR middle-end/46932
23449 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
23451 2017-07-26 Martin Liska <mliska@suse.cz>
23454 * function.c (expand_function_start): Make expansion of
23455 nonlocal_goto_save_area after parm_birth_insn.
23457 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23459 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
23460 from all CPU target flags enable members.
23462 2017-07-26 Richard Biener <rguenther@suse.de>
23464 * genmatch.c (dt_simplify::gen): Make iterator vars const.
23465 (decision_tree::gen): Make 'type' const.
23466 (write_predicate): Likewise.
23468 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23470 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
23471 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
23472 (rs6000_option_override_internal): Likewise.
23473 (rs6000_expand_vector_set): Likewise.
23474 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
23475 (TARGET_UPPER_REGS_SF): Likewise.
23476 (TARGET_UPPER_REGS_DI): Likewise.
23477 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
23478 (TARGET_DIRECT_MOVE_64BIT): Likewise.
23479 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
23480 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23481 (Splitters for DI constants in Altivec registers): Likewise.
23482 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
23483 (vsx_set_v4sf_p9): Likewise.
23484 (vsx_set_v4sf_p9_zero): Likewise.
23485 (vsx_insert_extract_v4sf_p9): Likewise.
23486 (vsx_insert_extract_v4sf_p9_2): Likewise.
23488 2017-07-25 Carl Love <cel@us.ibm.com>
23490 * doc/extend.texi: Update the built-in documentation file for the
23491 existing built-in functions
23492 vector signed char vec_cnttz (vector signed char);
23493 vector unsigned char vec_cnttz (vector unsigned char);
23494 vector signed short vec_cnttz (vector signed short);
23495 vector unsigned short vec_cnttz (vector unsigned short);
23496 vector signed int vec_cnttz (vector signed int);
23497 vector unsigned int vec_cnttz (vector unsigned int);
23498 vector signed long long vec_cnttz (vector signed long long);
23499 vector unsigned long long vec_cnttz (vector unsigned long long);
23501 2017-07-25 Andrew Pinski <apinski@cavium.com>
23503 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
23504 accesses where the use is for the first operand of a BIT_INSERT.
23506 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23509 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23510 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23512 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23514 * config/i386/gstabs.h: Delete.
23515 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
23517 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
23519 * config/i386/i386.c (ix86_decompose_address): Do not check for
23520 register RTX when looking at index_reg or base_reg.
23521 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
23523 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
23525 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
23526 to update EH info here.
23528 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23530 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
23532 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23534 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
23536 2017-07-25 Torsten Duwe <duwe@suse.de>
23538 * common.opt: Introduce -fpatchable-function-entry
23539 command line option, and its variables function_entry_patch_area_size
23540 and function_entry_patch_area_start.
23541 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
23542 including a two-value parser.
23543 * target.def (print_patchable_function_entry): New target hook.
23544 * targhooks.h (default_print_patchable_function_entry): New function.
23545 * targhooks.c (default_print_patchable_function_entry): Likewise.
23546 * toplev.c (process_options): Switch off IPA-RA if
23547 patchable function entries are being generated.
23548 * varasm.c (assemble_start_function): Look at the
23549 patchable-function-entry command line switch and current
23550 function attributes and maybe generate NOP instructions by
23551 calling the print_patchable_function_entry hook.
23552 * doc/extend.texi: Document patchable_function_entry attribute.
23553 * doc/invoke.texi: Document -fpatchable_function_entry
23554 command line option.
23555 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
23557 * doc/tm.texi: Re-generate.
23559 2017-07-25 Jakub Jelinek <jakub@redhat.com>
23562 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
23563 TARGET_AVX512DQ rather than TARGET_AVX512BW.
23565 2017-07-25 Tamar Christina <tamar.christina@arm.com>
23567 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
23569 2017-07-25 Richard Biener <rguenther@suse.de>
23571 PR tree-optimization/81455
23572 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
23573 not walk in cycles when looking for guards.
23575 2017-07-25 Richard Biener <rguenther@suse.de>
23577 PR tree-optimization/81529
23578 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
23579 when optimizing backedge uses.
23581 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
23583 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
23585 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
23586 to dl_section_ref. On AIX, append an expression to subtract
23587 the size of the section length to dl_section_ref.
23589 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
23591 * configure.ac: If any of the config.* scripts fail, exit 1.
23592 * configure: Regenerate.
23594 2017-07-25 Richard Biener <rguenther@suse.de>
23596 PR middle-end/81546
23597 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
23598 of immediate uses, be more verbose on errors.
23600 2017-07-25 Richard Biener <rguenther@suse.de>
23602 PR tree-optimization/81510
23603 * tree-vect-loop.c (vect_is_simple_reduction): When the
23604 reduction stmt is not inside the loop bail out.
23606 2017-07-25 Richard Biener <rguenther@suse.de>
23608 PR tree-optimization/81303
23609 * tree-vect-loop-manip.c (vect_loop_versioning): Build
23610 profitability check against LOOP_VINFO_NITERSM1.
23612 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23614 * domwalk.c (cmp_bb_postorder): Simplify.
23615 (sort_bbs_postorder): New function. Use it...
23616 (dom_walker::walk): ...here to optimize common cases.
23618 2017-07-25 Martin Liska <mliska@suse.cz>
23621 * ipa-visibility.c (function_and_variable_visibility): Make the
23622 redirection just on target that supports aliasing.
23623 Fix GNU coding style.
23625 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
23628 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
23630 * config/arm/rtems.h: Likewise.
23631 * config/bfin/rtems.h: Likewise.
23632 * config/i386/rtemself.h: Likewise.
23633 * config/lm32/rtems.h: Likewise.
23634 * config/m32c/rtems.h: Likewise.
23635 * config/m68k/rtemself.h: Likewise.
23636 * config/microblaze/rtems.h: Likewise.
23637 * config/mips/rtems.h: Likewise.
23638 * config/moxie/rtems.h: Likewise.
23639 * config/nios2/rtems.h: Likewise.
23640 * config/powerpcspe/rtems.h: Likewise.
23641 * config/rs6000/rtems.h: Likewise.
23642 * config/rtems.h: Likewise.
23643 * config/sh/rtems.h: Likewise.
23644 * config/sh/rtemself.h: Likewise.
23645 * config/sparc/rtemself.h: Likewise.
23647 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23650 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
23651 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
23652 * tree-ssa-structalias.c (alias_get_name): Same.
23654 2017-07-25 Bin Cheng <bin.cheng@arm.com>
23657 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
23658 instructions if no du chain is found.
23660 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23662 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
23664 2017-07-25 Richard Biener <rguenther@suse.de>
23666 PR middle-end/81505
23667 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
23670 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23672 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
23673 upper-regs options.
23674 (ISA_2_7_MASKS_SERVER): Likewise.
23675 (ISA_3_0_MASKS_IEEE): Likewise.
23676 (OTHER_P8_VECTOR_MASKS): Likewise.
23677 (OTHER_VSX_VECTOR_MASKS): Likewise.
23678 (POWERPC_MASKS): Likewise.
23679 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
23680 duplicate list of options.
23681 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
23682 explicit -mupper-regs options.
23683 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
23684 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
23685 alias for -mupper-regs-df.
23686 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
23687 (rs6000_init_hard_regno_mode_ok): Likewise.
23688 (rs6000_option_override_internal): Likewise.
23689 (rs6000_opt_masks): Likewise.
23690 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
23691 options in terms of whether -mvsx or -mpower8-vector was used.
23692 (TARGET_UPPER_REGS_DI): Likewise.
23693 (TARGET_UPPER_REGS_SF): Likewise.
23694 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
23695 -mupper-regs-* options.
23697 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23699 * passes.c (emergency_dump_function): Print some empty lines and a
23700 header before the RTL dump.
23702 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23704 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
23706 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
23709 * config/aarch64/aarch64.c (aarch64_classify_symbol):
23710 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
23712 2017-07-24 Carl Love <cel@us.ibm.com>
23714 * config/rs6000/rs6000-c.c: Add support for built-in functions
23715 vector float vec_extract_fp32_from_shorth (vector unsigned short);
23716 vector float vec_extract_fp32_from_shortl (vector unsigned short);
23717 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
23718 vec_extract_fp_from_shortl): Add defines for the two builtins.
23719 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
23720 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
23722 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
23723 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
23724 * doc/extend.texi: Update the built-in documentation file for the
23725 new built-in function.
23727 2017-07-24 Jakub Jelinek <jakub@redhat.com>
23730 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
23732 * doc/generic.texi: Likewise.
23733 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23734 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23736 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
23738 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
23739 (aarch64_mls_elt_merge<mode>): Likewise.
23741 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
23743 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23744 having __cxa_atexit.
23746 2017-07-23 Michael Collison <michael.collison@arm.com>
23748 * config/arm/arm.c (arm_option_override): Deprecate
23749 use of -mstructure-size-boundary.
23750 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
23751 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
23753 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23756 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23757 Reduce cost estimate for direct moves.
23759 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
23762 * config/i386/i386.c (ix86_option_override_internal): Disable
23763 BMI, BMI2 and TBM instructions for -m16.
23765 2017-07-21 Carl Love <cel@us.ibm.com>
23767 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23768 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23769 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23770 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23771 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23772 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23773 VMULOSW): New enum "unspec" values.
23774 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23775 altivec_vmulosw): New patterns.
23776 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23777 VMULOSW): Add definitions.
23779 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
23781 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
23782 (qdf24xx): Likewise.
23783 * config/aarch64/aarch64-options-extensions.def (rdma); New.
23784 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
23785 (AARCH64_FL_V8_1): Renumber.
23786 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
23787 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
23788 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
23789 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
23790 rdma to feature modifiers list.
23792 2017-07-21 Yury Gribov <tetra2005@gmail.com>
23794 PR middle-end/56727
23795 * ipa-visibility (function_and_variable_visibility): Convert
23796 recursive PLT call to direct call if appropriate.
23798 2017-07-21 Andrew Pinski <apinski@cavium.com>
23800 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
23801 operand 1 to see if the types precision matches.
23802 * fold-const.c (operand_equal_p): Likewise.
23804 2017-07-21 Richard Biener <rguenther@suse.de>
23806 PR tree-optimization/81303
23807 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23808 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23809 (vect_peeling_hash_get_lowest_cost): Adjust.
23810 (vect_enhance_data_refs_alignment): Likewise. Use
23811 vect_get_peeling_costs_all_drs to compute the penalty for no
23812 peeling to match up costs.
23814 2017-07-21 Richard Biener <rguenther@suse.de>
23816 PR tree-optimization/81500
23817 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23818 we didn't identify a reduction path.
23820 2017-07-21 Tom de Vries <tom@codesourcery.com>
23821 Cesar Philippidis <cesar@codesourcery.com>
23823 PR gcov-profile/81442
23824 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23827 2017-07-21 Tom de Vries <tom@codesourcery.com>
23830 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23832 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23833 nvptx_override_options_after_change.
23835 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23837 * dwarf2out.c (output_file_names): Avoid double testing for
23838 dwarf_version >= 5.
23840 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23842 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
23844 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23846 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23848 (try_crossjump_to_edge): Do not punt on partitioned functions.
23850 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23852 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23853 Put all BBs reachable only via paths crossing cold region to cold
23855 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23857 2016-07-21 Richard Biener <rguenther@suse.de>
23859 PR tree-optimization/81303
23860 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23861 into account prologue and epilogue iterations when raising
23862 min_profitable_iters to sth at least covering one vector iteration.
23864 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23866 * config/arm/arm.c (arm_test_cpu_arch_dat):
23869 2017-07-20 Nathan Sidwell <nathan@acm.org>
23871 Remove TYPE_METHODS.
23872 * tree.h (TYPE_METHODS): Delete.
23873 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23874 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23875 (dbxout_type_methods): Scan TYPE_FIELDS.
23876 (dbxout_type): Don't check TYPE_METHODS here.
23877 * function.c (use_register_for_decl): Always ignore register for
23878 class types when not optimizing.
23879 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23880 * tree.c (free_lang_data_in_type): Stitch out member functions and
23881 templates from TYPE_FIELDS.
23882 (build_distinct_type_copy, verify_type_variant,
23883 verify_type): Member fns are on TYPE_FIELDS.
23884 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23885 * tree-pretty-print.c (dump_generic_node): Likewise.
23887 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23890 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23891 V2TImode and V4TImode.
23892 (ix86_expand_vector_extract): Likewise.
23893 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23894 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23895 (ssescalarmode): Handle V4TImode and V2TImode.
23896 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23897 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23898 (VEXTRACTI128_MODE): New mode iterator.
23899 (splitter for *vec_extractv?ti first element): New.
23900 (VEC_INIT_MODE): New mode iterator.
23901 (vec_init<mode>): Consolidate 3 expanders into one using
23902 VEC_INIT_MODE mode iterator.
23904 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23906 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23907 non_spilled_static_chain_regno_p.
23909 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23911 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
23913 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
23915 * bb-reorder.c (connect_traces): Allow copying of blocks within
23918 2017-07-20 Richard Biener <rguenther@suse.de>
23920 * gimple.h (gimple_phi_result): Add gphi * overload.
23921 (gimple_phi_result_ptr): Likewise.
23922 (gimple_phi_arg): Likewise. Adjust index assert to only
23923 allow actual argument accesses rather than all slots available
23925 (gimple_phi_arg_def): Add gphi * overload.
23926 * tree-phinodes.c (make_phi_node): Initialize only actual
23928 (resize_phi_node): Clear memory not covered by old node,
23929 do not initialize excess argument slots.
23930 (reserve_phi_args_for_new_edge): Initialize new argument slot
23933 2017-07-20 Bin Cheng <bin.cheng@arm.com>
23935 PR tree-optimization/81388
23937 2016-07-21 Bin Cheng <bin.cheng@arm.com>
23939 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
23940 by removing computation of may_be_zero.
23942 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23943 Tom de Vries <tom@codesourcery.com>
23945 PR middle-end/81030
23946 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
23947 when gimple level profile disagrees with what RTL expander did.
23949 2017-07-20 Richard Biener <rguenther@suse.de>
23951 PR tree-optimization/61171
23952 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
23953 (vect_analyze_stmt): Add slp instance parameter.
23954 (vectorizable_reduction): Likewise.
23955 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
23956 (vect_is_simple_reduction): Deal with chains not detected
23957 as SLP reduction chain, specifically not properly associated
23958 chains containing a mix of plus/minus.
23959 (get_reduction_op): Remove.
23960 (get_initial_defs_for_reduction): Simplify, pass in whether
23961 this is a reduction chain, pass in the SLP node for the PHIs.
23962 (vect_create_epilog_for_reduction): Get the SLP instance as
23964 (vectorizable_reduction): Get the SLP instance as arg.
23965 During analysis remember the SLP node with the PHIs in the
23966 instance. Simplify getting at the vectorized reduction PHIs.
23967 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
23968 through SLP instance.
23969 (vect_slp_analyze_operations): Likewise.
23970 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
23971 (vect_transform_stmt): Likewise.
23973 2017-07-20 Tom de Vries <tom@codesourcery.com>
23975 PR tree-optimization/81489
23976 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
23977 read of phi arg location to before loop that modifies phi.
23979 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23981 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
23984 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23986 PR middle-end/81331
23987 * except.c (execute): Fix ordering issue.
23989 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23991 PR rtl-optimization/81423
23992 * combine.c (make_compound_operation_int): Don't try to optimize
23993 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
23995 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
23997 PR rtl-optimization/81423
23998 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
23999 with a constant that is -1 in the truncated to mode.
24001 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
24003 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
24004 (determine_unlikely_bbs): ... here.
24005 * predict.h (propagate_unlikely_bbs_forward): Declare.
24006 * cfgexpand.c (pass_expand::execute): Use it.
24007 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
24009 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
24010 propagate_unlikely_bbs_forward.
24012 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
24014 PR middle-end/81331
24015 * except.c (maybe_add_nop_after_section_switch): New function.
24018 2017-07-19 Tom de Vries <tom@codesourcery.com>
24020 * gimple.h (gimple_phi_set_arg): Make assert more strict.
24022 2017-07-19 Tom de Vries <tom@codesourcery.com>
24024 * gimple.h (gimple_phi_arg): Make assert more strict.
24026 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
24028 * config.gcc (powerpc*-*-*): Add mmintrin.h.
24029 * config/rs6000/mmintrin.h: New file.
24030 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
24032 2017-07-19 Jakub Jelinek <jakub@redhat.com>
24034 PR tree-optimization/81346
24035 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
24037 2017-07-19 Tom de Vries <tom@codesourcery.com>
24039 * config/nvptx/nvptx.md (VECIM): Add V2DI.
24041 2017-07-19 Tom de Vries <tom@codesourcery.com>
24043 * config/nvptx/nvptx-modes.def: Add V2DImode.
24044 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
24045 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
24046 (nvptx_output_mov_insn): Handle lack of mov.b128.
24047 (nvptx_print_operand): Handle 'H' and 'L' codes.
24048 (nvptx_vector_mode_supported): Allow V2DImode.
24049 (nvptx_preferred_simd_mode): New function.
24050 (nvptx_data_alignment): New function.
24051 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
24052 nvptx_preferred_simd_mode.
24053 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
24055 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
24057 2017-07-19 Tom de Vries <tom@codesourcery.com>
24059 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
24060 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
24061 (nvptx_vector_mode_supported): New function. Allow V2SImode.
24062 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
24063 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
24064 (mov<VECIM>_insn): New define_insn.
24065 (define_expand "mov<VECIM>): New define_expand.
24067 2017-07-19 Tom de Vries <tom@codesourcery.com>
24069 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
24071 2017-07-19 Jakub Jelinek <jakub@redhat.com>
24073 PR tree-optimization/81346
24074 * fold-const.h (fold_div_compare, range_check_type): Declare.
24075 * fold-const.c (range_check_type): New function.
24076 (build_range_check): Use range_check_type.
24077 (fold_div_compare): No longer static, rewritten into
24078 a match.pd helper function.
24079 (fold_comparison): Don't call fold_div_compare here.
24080 * match.pd (X / C1 op C2): New optimization using fold_div_compare
24081 as helper function.
24083 2017-07-19 Nathan Sidwell <nathan@acm.org>
24085 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
24086 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
24087 * tree.c (find_decls_types_r, verify_type): Use
24088 TYPE_{MIN,MAX}_VALUE_RAW.
24089 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
24090 (hash_tree): Likewise.
24091 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
24093 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
24096 2017-07-18 Tom de Vries <tom@codesourcery.com>
24098 PR middle-end/81464
24099 * omp-expand.c (expand_omp_for_static_chunk): Handle
24100 equal-argument loop exit phi.
24102 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
24105 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
24106 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
24107 operand 2 predicate.
24108 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
24109 operand 2 predicate.
24110 (ror,rol -> rorx splitters): Use const_int_operand as
24111 operand 2 predicate.
24113 2017-06-18 Richard Biener <rguenther@suse.de>
24115 PR tree-optimization/81410
24116 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
24117 the gap in the ! slp_perm SLP case after each group.
24119 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24121 PR middle-end/81463
24122 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
24125 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24127 PR middle-end/81462
24128 * predict.c (set_even_probabilities): Cleanup; do not affect
24129 probabilities that are already known.
24130 (combine_predictions_for_bb): Call even when count is set.
24132 2017-07-18 Nathan Sidwell <nathan@acm.org>
24134 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
24137 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24140 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
24141 optimization for loop niter analysis.
24143 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
24146 * config/avr/avr.c (avr_optimize_casesi): Don't use
24147 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
24149 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
24151 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
24152 body_cost_vec from _vect_peel_extended_info.
24153 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
24154 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
24157 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24159 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
24161 2017-07-18 Richard Biener <rguenther@suse.de>
24163 PR tree-optimization/80620
24164 PR tree-optimization/81403
24165 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
24166 info when re-using a VN table entry.
24168 2017-07-18 Richard Biener <rguenther@suse.de>
24170 PR tree-optimization/81418
24171 * tree-vect-loop.c (vectorizable_reduction): Properly compute
24172 vectype_in. Verify that with lane-reducing reduction operations
24173 we have a single def-use cycle.
24175 2017-07-17 Carl Love <cel@us.ibm.com>
24177 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
24179 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24180 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24181 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24182 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24183 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24184 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24185 VMULOSW): New enum "unspec" values.
24186 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24187 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24188 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24189 altivec_vmulosw): New patterns.
24190 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24191 VMULOSW): Add definitions.
24193 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
24195 * config/alpha/alpha.c: Include predict.h.
24197 2017-07-17 Yury Gribov <tetra2005@gmail.com>
24199 * tree-vrp.c (compare_assert_loc): Fix comparison function
24200 to return predictable results.
24202 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24204 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
24206 (subdi3): Likewise.
24207 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
24208 * doc/invoke.texi (mexpand-adddi): Update text.
24210 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24212 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
24213 that also clobbers the CC register. The old expand code is moved
24215 (*arc_clzsi2): ... here.
24216 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
24217 the CC register. The old expand code is moved to ...
24218 (arc_ctzsi2): ... here.
24220 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24222 * config/arc/arc.opt (mindexed-loads): Use initial value
24223 TARGET_INDEXED_LOADS_DEFAULT.
24224 (mauto-modify-reg): Use initial value
24225 TARGET_AUTO_MODIFY_REG_DEFAULT.
24226 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24227 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24228 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24229 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24231 2017-07-17 Martin Liska <mliska@suse.cz>
24234 * opts.c (finish_options): Do not allow -fgnu-tm
24235 w/ -fsanitize={kernel-,}address. Say sorry.
24237 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24240 * tree-loop-distribution.c (classify_partition): Only assert on
24241 numer of iterations.
24242 (merge_dep_scc_partitions): Delete prameter. Update function call.
24243 (distribute_loop): Remove code handling loop with unknown niters.
24244 (pass_loop_distribution::execute): Skip loop with unknown niters.
24246 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24249 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
24250 function sort_partitions_by_post_order.
24252 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24254 PR tree-optimization/81374
24255 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
24256 the max index of basic blocks, rather than number of basic blocks.
24258 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24260 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
24262 (arc_legitimate_pic_operand_p): Likewise.
24263 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
24265 (arc_needs_pcl_p): Likewise.
24266 (arc_legitimate_pc_offset_p): Likewise.
24267 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
24268 function is also used in constrains.md.
24269 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
24270 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
24271 PLUS. Only return true/false in known cases, otherwise assert.
24272 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
24273 is already called in arc_legitimate_constant_p.
24274 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
24276 (LEGITIMATE_PIC_OPERAND_P): Use
24277 arc_raw_symbolic_reference_mentioned_p function.
24278 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
24283 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24284 Andrew Burgess <andrew.burgess@embecosm.com>
24286 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
24287 (arc_return_address_register): New function.
24288 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
24289 (arc_handle_fndecl_attribute): Add naked attribute.
24290 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
24291 (TARGET_WARN_FUNC_RETURN): Likewise.
24292 (arc_allocate_stack_slots_for_args): New function.
24293 (arc_warn_func_return): Likewise.
24294 (machine_function): Change type fn_type.
24295 (arc_compute_function_type): Consider new naked function type,
24296 change function return type.
24297 (arc_must_save_register): Adapt to handle new
24298 arc_compute_function_type's return type.
24299 (arc_expand_prologue): Likewise.
24300 (arc_expand_epilogue): Likewise.
24301 (arc_return_address_regs): Delete.
24302 (arc_return_address_register): New function.
24303 (arc_epilogue_uses): Use above function.
24304 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
24305 (arc_function_type): Change encoding, add naked type.
24306 (ARC_INTERRUPT_P): Change to handle the new encoding.
24307 (ARC_FAST_INTERRUPT_P): Likewise.
24308 (ARC_NORMAL_P): Define.
24309 (ARC_NAKED_P): Likewise.
24310 (arc_compute_function_type): Delete prototype.
24311 * config/arc/arc.md (in_ret_delay_slot): Use
24312 arc_return_address_register function.
24313 (simple_return): Likewise.
24314 (p_return_i): Likewise.
24316 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24318 PR tree-optimization/81428
24319 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
24320 can't be built for those types.
24322 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24324 Remove stuff dead since r239246.
24326 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
24327 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
24328 (avr_inform_devices): Remove dead stuff.
24330 2017-07-17 Tamar Christina <tamar.christina@arm.com>
24332 * config/arm/arm_neon.h: Fix softp typo.
24334 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24336 PR tree-optimization/81365
24337 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
24338 aggregate moves onto bb predecessor edges, make sure there are no
24339 loads that could alias the lhs in between the start of bb and the
24342 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24345 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
24346 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
24347 [LSHIFTRT, outer_code = TRUNCATE]: Same.
24349 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24351 PR tree-optimization/81396
24352 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
24353 (init_symbolic_number): Initialize it to 1.
24354 (perform_symbolic_merge): Add n_ops from both operands into the new
24356 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
24357 without base_addr as useless if they need more than one operation.
24358 (bswap_replace): Handle !bswap case for NULL base_addr.
24360 2017-07-17 Tom de Vries <tom@codesourcery.com>
24363 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
24366 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
24368 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
24369 conditional builtin define __FIX_LEON3FT_B2BST.
24371 2017-07-17 Daniel Cederman <cederman@gaisler.com>
24373 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
24374 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
24377 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24379 PR rtl-optimization/81424
24380 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
24381 to remove potential trapping from operands if -fnon-call-exceptions.
24383 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24385 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
24386 profile_proability for scalling.
24387 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
24389 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24391 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
24393 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24395 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
24396 fixpoint arithmetics.
24398 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24400 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
24401 fixpoint arithmetics.
24403 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24405 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
24406 fixpoint arithmetics.
24408 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24410 * profile-count.h (profile_probability::from_reg_br_prob_note,
24411 profile_probability::to_reg_br_prob_note): New functions.
24412 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
24413 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
24414 * predict.c (probability_reliable_p): Update.
24415 (edge_probability_reliable_p): Update.
24416 (br_prob_note_reliable_p): Update.
24417 (invert_br_probabilities): Update.
24418 (add_reg_br_prob_note): New function.
24419 (combine_predictions_for_insn): Update.
24420 * asan.c (asan_clear_shadow): Update.
24421 * cfgbuild.c (compute_outgoing_frequencies): Update.
24422 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24423 (update_br_prob_note): Update.
24424 (rtl_verify_edges): Update.
24425 (purge_dead_edges): Update.
24426 (fixup_reorder_chain): Update.
24427 * emit-rtl.c (try_split): Update.
24428 * ifcvt.c (cond_exec_process_insns): Update.
24429 (cond_exec_process_if_block): Update.
24430 (dead_or_predicable): Update.
24431 * internal-fn.c (expand_addsub_overflow): Update.
24432 (expand_neg_overflow): Update.
24433 (expand_mul_overflow): Update.
24434 * loop-doloop.c (doloop_modify): Update.
24435 * loop-unroll.c (compare_and_jump_seq): Update.
24436 * optabs.c (emit_cmp_and_jump_insn_1): Update.
24437 * predict.h: Update.
24438 * reorg.c (mostly_true_jump): Update.
24440 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
24441 * config/alpha/alpha.c (emit_unlikely_jump): Update.
24442 * config/arc/arc.c: (emit_unlikely_jump): Update.
24443 * config/arm/arm.c: (emit_unlikely_jump): Update.
24444 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
24445 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
24446 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
24447 (ix86_print_operand): Update.
24448 (ix86_split_fp_branch): Update.
24449 (predict_jump): Update.
24450 * config/ia64/ia64.c (ia64_print_operand): Update.
24451 * config/mmix/mmix.c (mmix_print_operand): Update.
24452 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
24453 (rs6000_expand_split_stack_prologue): Update.
24454 * config/rs6000/rs6000.c: Update.
24455 * config/s390/s390.c (s390_expand_vec_strlen): Update.
24456 (s390_expand_vec_movstr): Update.
24457 (s390_expand_cs_tdsi): Update.
24458 (s390_expand_split_stack_prologue): Update.
24459 * config/sh/sh.c (sh_print_operand): Update.
24460 (expand_cbranchsi4): Update.
24461 (expand_cbranchdi4): Update.
24462 * config/sparc/sparc.c (output_v9branch): Update.
24463 * config/spu/spu.c (get_branch_target): Update.
24464 (ea_load_store_inline): Update.
24465 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
24466 * config/tilepro/tilepro.c: Update.
24468 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24470 * gimplify.c (mostly_copy_tree_r): Revert latest change.
24471 (gimplify_save_expr): Likewise.
24473 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24475 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
24477 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24479 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
24481 * timevar.def (TV_IPA_FNSUMMARY): Define.
24483 2017-07-16 Daniel Cederman <cederman@gaisler.com>
24485 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
24486 to back store errata sensitive sequence from being generated.
24487 (sqrtdf2_fix): Likewise.
24489 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24491 * tree-ssa-threadupdate.c (compute_path_counts,
24492 update_joiner_offpath_counts): Use profile_probability.
24494 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
24497 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24499 * config/arm/arm-c.c (arm_cpu_builtins): Define
24500 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24502 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
24504 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24505 array entries to represent __ieee128 versions of the
24506 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
24507 scalar_extract_sig, and scalar_insert_exp built-in functions.
24508 (altivec_resolve_overloaded_builtin): Add special case handling
24509 for the __builtin_scalar_insert_exp function, as represented by
24510 the P9V_BUILTIN_VEC_VSIEDP constant.
24511 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
24512 exponent support for __ieee128 argument.
24513 (VSESQP): Add scalar extract signature support for __ieee128
24515 (VSTDCNQP): Add scalar test negative support for __ieee128
24517 (VSIEQP): Add scalar insert exponent support for __int128 argument
24518 with __ieee128 result.
24519 (VSIEQPF): Add scalar insert exponent support for __ieee128
24520 argument with __ieee128 result.
24521 (VSTDCQP): Add scalar test data class support for __ieee128
24523 (VSTDCNQP): Add overload support for scalar test negative with
24524 __ieee128 argument.
24525 (VSTDCQP): Add overload support for scalar test data class
24526 __ieee128 argument.
24527 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
24528 UNSPEC_VSX_SXSIGDP.
24529 (UNSPEC_VSX_SIEXPQP): New constant.
24530 (xsxexpqp): New insn for VSX scalar extract exponent quad
24532 (xsxsigqp): New insn for VSX scalar extract significand quad
24534 (xsiexpqpf): New insn for VSX scalar insert exponent quad
24535 precision with floating point argument.
24536 (xststdcqp): New expand for VSX scalar test data class quad
24538 (xststdcnegqp): New expand for VSX scalar test negative quad
24540 (xststdcqp): New insn to match expansions for VSX scalar test data
24541 class quad precision and VSX scalar test negative quad precision.
24542 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
24543 special case operand checking to enforce that second operand of
24544 VSX scalar test data class with quad precision argument is a 7-bit
24546 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24547 prototypes and descriptions of __ieee128 versions of
24548 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
24549 scalar_test_data_class, and scalar_test_neg built-in functions.
24551 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24553 PR tree-optimization/81162
24554 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
24555 replace a negate with an add.
24557 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
24559 * doc/invoke.texi (arm/-mcpu): Document +crypto.
24561 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24563 * config/arm/arm-c.c (arm_cpu_builtins): Define
24564 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24566 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24568 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
24569 (armv8-r): Set ARM Cortex-R52 as default CPU.
24570 * config/arm/arm-tables.opt: Regenerate.
24571 * config/arm/arm-tune.md: Regenerate.
24572 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24574 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
24575 extension for -mcpu=cortex-r52.
24577 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24579 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
24580 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
24581 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
24582 (fp-armv8): Define it as FP_ARMv8 only.
24583 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
24584 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
24586 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
24587 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
24588 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
24589 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
24590 than TARGET_FPU_ARMV8.
24591 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
24592 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
24593 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
24595 * config/arm/neon.md (neon_vrint): Likewise.
24596 (neon_vcvt): Likewise.
24597 (neon_<fmaxmin_op><mode>): Likewise.
24598 (<fmaxmin><mode>3): Likewise.
24599 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
24600 * config/arm/predicates.md (arm_cond_move_operator): Check against
24601 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
24603 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
24605 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
24606 to top of function.
24608 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24610 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
24611 loop in comment with memset.
24613 2017-07-14 Martin Liska <mliska@suse.cz>
24615 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
24616 * dwarf2out.c (is_java): Remove the function.
24617 (output_pubname): Remove usage of the function.
24618 (lower_bound_default): Remove usage of DW_LANG_Java.
24619 (gen_compile_unit_die): Likewise.
24620 * gcc.c: Remove compiler defaults for .java and .zip files.
24621 * gimple-expr.c (remove_suffix): Change as there's no longer
24622 extension than 4-letter one.
24623 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
24624 (gimplify_save_expr): Likewise.
24625 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
24626 as it's possible even for other languages than Java.
24627 * langhooks.h (struct lang_hooks): Remove Java from a comment.
24628 * lto-opts.c (lto_write_options): Remove reference to Java.
24629 * opts.c (strip_off_ending): Update file extension handling.
24630 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
24631 * tree-eh.c (lower_resx): Likewise.
24632 * tree.c (free_lang_data_in_type): Remove dead code.
24633 (find_decls_types_r): Likewise.
24634 (build_common_builtin_nodes): Remove Java from a comment.
24635 (verify_type): Remove dead code.
24636 * varasm.c (assemble_external): Remove Java from a comment.
24638 2017-07-14 Martin Liska <mliska@suse.cz>
24640 * opts.c (finish_options): Add quotes.
24641 (common_handle_option): Likewise.
24643 2017-07-14 Martin Liska <mliska@suse.cz>
24645 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
24646 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
24647 Remove N_SO_PASCAL.
24648 * dwarf2out.c (lower_bound_default): Do not handle
24650 (gen_compile_unit_die): Likewise.
24651 * gcc.c: Remove default extension binding for GNU Pascal.
24652 * stmt.c: Remove Pascal language from a comment.
24653 * xcoffout.c: Likewise.
24655 2017-07-13 David Malcolm <dmalcolm@redhat.com>
24658 * diagnostic-show-locus.c (fixit_cmp): New function.
24659 (layout::layout): Sort m_fixit_hints.
24660 (column_range::column_range): Assert that the values are valid.
24661 (struct char_span): New struct.
24662 (correction::overwrite): New method.
24663 (struct source_line): New struct.
24664 (line_corrections::add_hint): Add assertions. Reimplement memcpy
24665 calls in terms of classes source_line and char_span, and
24666 correction::overwrite.
24667 (selftest::test_overlapped_fixit_printing_2): New function.
24668 (selftest::diagnostic_show_locus_c_tests): Call it.
24670 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
24672 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
24673 early if there is no lhs.
24675 2017-07-13 Martin Liska <mliska@suse.cz>
24677 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
24678 (gen_reference_type_die): Likewise.
24679 * stor-layout.c: Remove Pascal-related comment.
24681 2017-07-13 Martin Liska <mliska@suse.cz>
24683 * opts.c (finish_options): Add quotes to error messages.
24684 (parse_sanitizer_options): Likewise.
24686 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
24688 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
24690 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
24692 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
24694 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24696 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
24698 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
24700 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24703 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
24704 provides the hardware capability bits, define the macro
24705 __BUILTIN_CPU_SUPPORTS__.
24706 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
24707 if GLIBC does not provide the hardware capability bits. Add a
24708 gcc_unreachable call if the built-in cpu function is neither
24709 __builtin_cpu_is nor __builtin_cpu_supports.
24710 (rs6000_get_function_versions_dispatcher): Change the warning
24711 that an old GLIBC is used which does not export the capability
24712 bits to be an error.
24713 * doc/extend.texi (target_clones attribute): Document the
24714 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
24715 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
24716 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
24717 the macros defined by GCC if the newer GLIBC is available.
24719 2017-07-12 Jeff Law <law@redhat.com>
24721 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
24722 remaining includes slightly.
24723 * config/riscv/riscv-builtins.c: Include profile-count.h.
24725 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24728 * config/avr/avr.c (avr_set_current_function): In diagnostic
24729 messages: Quote keywords and (parts of) identifiers.
24730 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
24733 2017-07-12 Carl Love <cel@us.ibm.com>
24735 * config/rs6000/rs6000-c.c: Add support for built-in functions
24736 vector bool char vec_revb (vector bool char);
24737 vector bool short vec_revb (vector short char);
24738 vector bool int vec_revb (vector bool int);
24739 vector bool long long vec_revb (vector bool long long);
24740 * doc/extend.texi: Update the built-in documentation file for the
24741 new built-in functions.
24743 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24745 * config/s390/s390.md: Remove movcc splitter.
24747 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24749 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
24750 load/store on condition.
24752 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24755 * config/avr/avr.c (avr_encode_section_info)
24756 [progmem && !TREE_READONLY]: Error if progmem object needs
24759 2017-07-11 Michael Collison <michael.collison@arm.com>
24761 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
24764 2017-07-11 Carl Love <cel@us.ibm.com>
24766 * config/rs6000/rs6000-c.c: Add support for builtins
24767 vector unsigned int vec_parity_lsbb (vector signed int);
24768 vector unsigned int vec_parity_lsbb (vector unsigned int);
24769 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
24770 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
24771 vector unsigned long long vec_parity_lsbb (vector signed long long);
24772 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
24773 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
24774 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
24775 * doc/extend.texi: Update the built-in documentation file for the
24776 new built-in functions.
24778 2017-07-11 David Malcolm <dmalcolm@redhat.com>
24780 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
24781 (layout::m_primary_loc): New field.
24782 (layout::layout): Initialize new field. Move location filtering
24783 logic from here to...
24784 (layout::maybe_add_location_range): ...this new method. Add
24785 support for filtering to just the lines already specified by other
24787 (layout::will_show_line_p): New method.
24788 (gcc_rich_location::add_location_if_nearby): New method.
24789 (selftest::test_add_location_if_nearby): New test function.
24790 (selftest::diagnostic_show_locus_c_tests): Call it.
24791 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
24794 2017-07-11 Tom de Vries <tom@codesourcery.com>
24796 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
24797 (bb_first_real_insn): New function.
24798 (nvptx_single): Add extra initialization of broadcasted condition
24801 2017-07-11 Nathan Sidwell <nathan@acm.org>
24803 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
24805 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24807 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24808 Binutils doc as TEXI will mess them up.
24809 * doc/invoke.texi (AVR Options): Same here.
24811 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24813 * config/sparc/sparc.opt (mfix-ut700): New option.
24814 (mfix-gr712rc): Likewise.
24815 (sparc_fix_b2bst): New variable.
24816 * doc/invoke.texi (SPARC options): Document them.
24817 (ARM options): Fix warnings.
24818 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24819 instructions to prevent sequences that can trigger the store-store
24820 errata for certain LEON3FT processors.
24821 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24822 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24823 * config/sparc/sparc.md (fix_b2bst): New attribute.
24824 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24826 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24829 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24831 (*rsqrtsf2_sse): Ditto.
24833 (div<mode>3): Macroize insn from divdf3 and divsf3
24834 using MODEF mode iterator.
24836 2017-07-10 Martin Sebor <msebor@redhat.com>
24838 PR tree-optimization/80397
24839 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24840 instead of testing for equality to INTEGER_TYPE.
24842 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
24844 * config.gcc: Remove uclibc from arc target spec.
24846 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24848 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24850 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24853 * lto-wrapper.c (remove_option): New function.
24854 (merge_and_complain): Merge PIC/PIE options more realistically.
24856 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24858 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24862 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24863 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24864 * config.in: Regenerate.
24865 * configure: Regenerate.
24866 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24867 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24868 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24869 (TARGET_GASISR_PROLOGUES): ...target mask.
24870 * common/config/avr/avr-common.c
24871 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24872 Set -mgas-isr-prologues.
24873 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24874 INSERT_PASS_BEFORE for it.
24875 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24876 * config/avr/avr.c (avr_option_override)
24877 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24878 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24879 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24880 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24881 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24883 (make_avr_pass_pre_proep): New function.
24884 (emit_push_sfr) <treg>: Add argument to function and use it
24885 instead of TMP_REG.
24886 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24887 and set machine->gasisr.yes.
24888 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24889 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24890 __gcc_isr.n_pushed to .L__stack_usage.
24891 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24892 (avr_asm_final_postscan_insn): ...this new static function.
24893 * config/avr/avr.h (machine_function)
24894 <is_no_gccisr, use_L__stack_usage>: New fields.
24895 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24896 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24897 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24898 (gasisr, *gasisr): New expander and insn.
24899 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24900 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24901 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24903 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24905 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24908 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24910 Move jump-tables out of .text again.
24913 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
24914 (ASM_OUTPUT_ADDR_VEC): New function.
24915 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
24916 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
24917 INSN_ADDRESSes as asm comment.
24918 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
24919 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
24920 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
24921 * config/avr/avr.md (*tablejump): Adjust comment.
24922 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
24923 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
24925 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
24926 (avr_output_addr_vec): New proto.
24927 (avr_log_t) <insn_addresses>: New field.
24929 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
24932 * config/i386/i386.c (ix86_function_arg_advance): Set
24933 outgoing_args_on_stack to true if there are outgoing arguments
24935 (ix86_function_arg): Likewise.
24936 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
24937 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
24938 * config/i386/i386.h (machine_function): Add
24939 outgoing_args_on_stack.
24941 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
24943 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
24944 supporting pthreds.
24945 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
24947 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
24949 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
24950 (REAL_H): Remove $(MACHMODE_H).
24951 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
24953 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
24954 $(MACHMODE_H) and double-int.h.
24955 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
24957 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
24960 2017-07-07 Andrew Pinski <apinski@cavium.com>
24962 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
24963 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
24965 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
24967 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
24968 Add warning if GCC was not configured to link against a GLIBC that
24969 exports the hardware capability bits.
24970 (make_resolver_func): Make resolver function private and not a
24971 COMDAT function. Create the name with clone_function_name instead
24972 of make_unique_name.
24975 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
24976 correct operand in doing the split.
24978 2017-07-07 Carl Love <cel@us.ibm.com>
24980 * config/rs6000/rs6000-c: Add support for built-in function
24981 vector unsigned short vec_pack_to_short_fp32 (vector float,
24983 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
24984 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
24985 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
24986 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
24987 (convert_4f32_8i16): Add define_expand.
24988 * doc/extend.texi: Update the built-in documentation file for the
24989 new built-in function.
24991 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24993 * config/sparc/m8.md: New file.
24994 * config/sparc/sparc.md: Include m8.md.
24996 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
24998 * config/sparc/sparc.opt: New option -mvis4b.
24999 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
25000 (sparc_option_override): Handle VIS4B.
25001 (enum sparc_builtins): Define
25002 SPARC_BUILTIN_DICTUNPACK{8,16,32},
25003 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
25004 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
25005 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
25006 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
25007 (check_constant_argument): New function.
25008 (sparc_vis_init_builtins): Define builtins
25009 __builtin_vis_dictunpack{8,16,32},
25010 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
25011 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
25012 __builtin_vis_fpcmpde{8,16,32}shl and
25013 __builtin_vis_fpcmpur{8,16,32}shl.
25014 (sparc_expand_builtin): Check that the constant operands to
25015 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
25016 constant and in range.
25017 * config/sparc/sparc-c.c (sparc_target_macros): Handle
25019 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
25020 (SPARC_IMM5_P): Likewise.
25021 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
25022 (enabled): Handle vis4b.
25023 (UNSPEC_DICTUNPACK): New unspec.
25024 (UNSPEC_FPCMPSHL): Likewise.
25025 (UNSPEC_FPUCMPSHL): Likewise.
25026 (UNSPEC_FPCMPDESHL): Likewise.
25027 (UNSPEC_FPCMPURSHL): Likewise.
25028 (cpu_feature): New CPU feature `vis4b'.
25029 (dictunpack{8,16,32}): New insns.
25030 (FPCSMODE): New mode iterator.
25031 (fpcscond): New code iterator.
25032 (fpcsucond): Likewise.
25033 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
25034 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
25035 (fpcmpde{8,16,32}{si,di}shl): Likewise.
25036 (fpcmpur{8,16,32}{si,di}shl): Likewise.
25037 * config/sparc/constraints.md: Define constraints `q' for unsigned
25038 2-bit integer constants and `t' for unsigned 5-bit integer
25040 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
25042 (imm5_operand_dictunpack16): Likewise.
25043 (imm5_operand_dictunpack32): Likewise.
25044 (imm2_operand): Likewise.
25045 * doc/invoke.texi (SPARC Options): Document -mvis4b.
25046 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
25047 ditunpack* and fpcmp*shl builtins.
25049 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25051 * config.gcc: Handle m8 in --with-{cpu,tune} options.
25052 * config.in: Add HAVE_AS_SPARC6 define.
25053 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
25055 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
25057 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
25058 (CPP_CPU_SPEC): Handle m8.
25059 (ASM_CPU_SPEC): Likewise.
25060 * config/sparc/sparc-opts.h (enum processor_type): Add
25062 * config/sparc/sparc.c (m8_costs): New struct.
25063 (sparc_option_override): Handle TARGET_CPU_m8.
25064 (sparc32_initialize_trampoline): Likewise.
25065 (sparc64_initialize_trampoline): Likewise.
25066 (sparc_issue_rate): Likewise.
25067 (sparc_register_move_cost): Likewise.
25068 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
25069 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
25070 (ASM_CPU64_DEFAULT_SPEC): Likewise.
25071 (CPP_CPU_SPEC): Handle M8.
25072 (ASM_CPU_SPEC): Likewise.
25073 (AS_M8_FLAG): Define.
25074 * config/sparc/sparc.md: Add m8 to the cpu attribute.
25075 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
25076 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
25078 * configure: Regenerate.
25079 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
25082 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25084 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
25086 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
25087 ("*movdi_insn_sp32"): Do not set v3pipe.
25088 ("*movsi_insn"): Likewise.
25089 ("*movdi_insn_sp64"): Likewise.
25090 ("*movsf_insn"): Likewise.
25091 ("*movdf_insn_sp32"): Likewise.
25092 ("*movdf_insn_sp64"): Likewise.
25093 ("*zero_extendsidi2_insn_sp64"): Likewise.
25094 ("*sign_extendsidi2_insn"): Likewise.
25095 ("*mov<VM32:mode>_insn"): Likewise.
25096 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25097 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25098 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25099 ("<vlop:code><VL:mode>3"): Likewise.
25100 ("*not_<vlop:code><VL:mode>3"): Likewise.
25101 ("*nand<VL:mode>_vis"): Likewise.
25102 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
25103 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
25104 ("one_cmpl<VL:mode>2"): Likewise.
25105 ("faligndata<VM64:mode>_vis"): Likewise.
25106 ("alignaddrsi_vis"): Likewise.
25107 ("alignaddrdi_vis"): Likweise.
25108 ("alignaddrlsi_vis"): Likewise.
25109 ("alignaddrldi_vis"): Likewise.
25110 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25111 ("bmaskdi_vis"): Likewise.
25112 ("bmasksi_vis"): Likewise.
25113 ("bshuffle<VM64:mode>_vis"): Likewise.
25114 ("cmask8<P:mode>_vis"): Likewise.
25115 ("cmask16<P:mode>_vis"): Likewise.
25116 ("cmask32<P:mode>_vis"): Likewise.
25117 ("pdistn<P:mode>_vis"): Likewise.
25118 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25120 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25122 * config/sparc/sparc.md ("subtype"): New insn attribute.
25123 ("*wrgsr_sp64"): Set insn subtype.
25124 ("*rdgsr_sp64"): Likewise.
25125 ("alignaddrsi_vis"): Likewise.
25126 ("alignaddrdi_vis"): Likewise.
25127 ("alignaddrlsi_vis"): Likewise.
25128 ("alignaddrldi_vis"): Likewise.
25129 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25130 ("fexpand_vis"): Likewise.
25131 ("fpmerge_vis"): Likewise.
25132 ("faligndata<VM64:mode>_vis"): Likewise.
25133 ("bshuffle<VM64:mode>_vis"): Likewise.
25134 ("cmask8<P:mode>_vis"): Likewise.
25135 ("cmask16<P:mode>_vis"): Likewise.
25136 ("cmask32<P:mode>_vis"): Likewise.
25137 ("fchksm16_vis"): Likewise.
25138 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
25139 ("fmean16_vis"): Likewise.
25140 ("fp<plusminus_insn>64_vis"): Likewise.
25141 ("<plusminus_insn>v8qi3"): Likewise.
25142 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25143 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
25144 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
25145 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
25146 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
25147 ("*movqi_insn"): Likewise.
25148 ("*movhi_insn"): Likewise.
25149 ("*movsi_insn"): Likewise.
25150 ("movsi_pic_gotdata_op"): Likewise.
25151 ("*movdi_insn_sp32"): Likewise.
25152 ("*movdi_insn_sp64"): Likewise.
25153 ("movdi_pic_gotdata_op"): Likewise.
25154 ("*movsf_insn"): Likewise.
25155 ("*movdf_insn_sp32"): Likewise.
25156 ("*movdf_insn_sp64"): Likewise.
25157 ("*zero_extendhisi2_insn"): Likewise.
25158 ("*zero_extendqihi2_insn"): Likewise.
25159 ("*zero_extendqisi2_insn"): Likewise.
25160 ("*zero_extendqidi2_insn"): Likewise.
25161 ("*zero_extendhidi2_insn"): Likewise.
25162 ("*zero_extendsidi2_insn_sp64"): Likewise.
25163 ("ldfsr"): Likewise.
25164 ("prefetch_64"): Likewise.
25165 ("prefetch_32"): Likewise.
25166 ("tie_ld32"): Likewise.
25167 ("tie_ld64"): Likewise.
25168 ("*tldo_ldub_sp32"): Likewise.
25169 ("*tldo_ldub1_sp32"): Likewise.
25170 ("*tldo_ldub2_sp32"): Likewise.
25171 ("*tldo_ldub_sp64"): Likewise.
25172 ("*tldo_ldub1_sp64"): Likewise.
25173 ("*tldo_ldub2_sp64"): Likewise.
25174 ("*tldo_ldub3_sp64"): Likewise.
25175 ("*tldo_lduh_sp32"): Likewise.
25176 ("*tldo_lduh1_sp32"): Likewise.
25177 ("*tldo_lduh_sp64"): Likewise.
25178 ("*tldo_lduh1_sp64"): Likewise.
25179 ("*tldo_lduh2_sp64"): Likewise.
25180 ("*tldo_lduw_sp32"): Likewise.
25181 ("*tldo_lduw_sp64"): Likewise.
25182 ("*tldo_lduw1_sp64"): Likewise.
25183 ("*tldo_ldx_sp64"): Likewise.
25184 ("*mov<VM32:mode>_insn"): Likewise.
25185 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25186 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25188 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25190 * config/sparc/sparc.md ("type"): New insn type viscmp.
25191 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
25193 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
25194 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
25195 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25196 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
25198 ("n7_vis_logical_11cycle"): Likewise.
25199 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
25200 * config/sparc/niagara2.md ("niag3_vis": Likewise.
25201 * config/sparc/niagara.md ("niag_vis"): Likewise.
25202 * config/sparc/ultra3.md ("us3_fga"): Likewise.
25203 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
25205 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25207 * config/sparc/sparc.md: New instruction type `bmask'.
25208 (bmaskdi_vis): Use the `bmask' type.
25209 (bmasksi_vis): Likewise.
25210 * config/sparc/ultra3.md (us3_array): Likewise.
25211 * config/sparc/niagara7.md (n7_array): Likewise.
25212 * config/sparc/niagara4.md (n4_array): Likewise.
25213 * config/sparc/niagara2.md (niag2_vis): Likewise.
25214 (niag3_vis): Likewise.
25215 * config/sparc/niagara.md (niag_vis): Likewise.
25217 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25219 * ipa-comdats.c: Remove optimize check from gate.
25220 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
25221 for functions not optimized.
25222 (ipa_fn_summary_read): Skip optimize check.
25223 (ipa_fn_summary_write): Likewise.
25224 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
25226 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
25228 (can_inline_edge_p): Check flag_pcc_struct_return for match.
25229 (check_callers): Give up on caller which is not optimized.
25230 (inline_small_functions): Likewise.
25231 (ipa_inline): Do not give up when not optimizing.
25232 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
25233 away unoptimizes cdtors.
25234 (whole_program_function_and_variable_visibility): Do
25235 ipa_discover_readonly_nonaddressable_vars in LTO mode.
25236 * ipa.c (process_references): Do not check optimize.
25237 (symbol_table::remove_unreachable_nodes): Update optimize check.
25238 (set_writeonly_bit): Update optimize check.
25239 (pass_ipa_cdtor_merge::gate): Do not check optimize.
25240 (pass_ipa_single_use::gate): Remove.
25242 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25244 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
25245 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
25246 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
25247 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
25248 permute_load, permute_store, adjust_extract, adjust_splat,
25249 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
25250 replace_swap_with_copy, dump_swap_insn_table,
25251 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
25252 recombine_lvx_pattern, recombine_stvx_pattern,
25253 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
25254 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
25255 to file rs6000-p8swap.c.
25256 * config/rs6000/rs6000-p8swap.c: New file.
25257 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
25258 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
25259 and rs6000*-*-* targets.
25261 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25263 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
25265 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25267 * lto-wrapper.c (merge_and_complain): Do not merge
25268 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
25269 fsigned_zeros, ftrapping_math, fwrapv.
25270 (append_compiler_options): Do not track these options.
25271 (append_linker_options): Likewie
25273 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25275 * cgraphunit.c (cgraph_node::finalize_function): When
25276 !flag_toplevel_reorde set no_reorder flag.
25277 (varpool_node::finalize_decl): Likewise.
25278 (symbol_table::compile): Drop no toplevel reorder path.
25280 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25282 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
25283 edges; zero probability is not better than uninitialized.
25285 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25287 * asan.h (asan_sanitize_allocas_p): Declare.
25288 * asan.c (asan_sanitize_allocas_p): New function.
25289 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
25290 (handle_builtin_alloca): Likewise.
25291 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
25292 if !asan_sanitize_allocas_p.
25293 * params.def (asan-instrument-allocas): Add new option.
25294 * params.h (ASAN_PROTECT_ALLOCAS): Define.
25295 * opts.c (common_handle_option): Disable allocas sanitization for
25298 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25300 * asan.c: Include gimple-fold.h.
25301 (get_last_alloca_addr): New function.
25302 (handle_builtin_stackrestore): Likewise.
25303 (handle_builtin_alloca): Likewise.
25304 (asan_emit_allocas_unpoison): Likewise.
25305 (get_mem_refs_of_builtin_call): Add new parameter, remove const
25306 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
25307 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
25308 (instrument_builtin_call): Pass gimple iterator to
25309 get_mem_refs_of_builtin_call.
25310 (last_alloca_addr): New global.
25311 * asan.h (asan_emit_allocas_unpoison): Declare.
25312 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
25313 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
25314 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
25315 if function calls alloca.
25316 * gimple-fold.c (replace_call_with_value): Remove static keyword.
25317 * gimple-fold.h (replace_call_with_value): Declare.
25318 * internal-fn.c: Include asan.h.
25319 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
25320 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
25322 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25324 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
25325 (C_SELFTEST_FLAGS): New.
25326 (CPP_SELFTEST_FLAGS): New.
25327 (SELFTEST_DEPS): New, from deps of s-selftest.
25328 (C_SELFTEST_DEPS): New, from deps of s-selftest.
25329 (CPP_SELFTEST_DEPS): New.
25330 (selftest): Add dependency on s-selftest-c++.
25331 (s-selftest): Rename to...
25332 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
25333 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
25334 than SELFTEST_FLAGS.
25335 (selftest-gdb): Rename to...
25336 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
25338 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
25339 (selftest-valgrind): Rename to...
25340 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
25342 (selftest-valgrind): Reintroduce as an alias for
25343 selftest-c-valgrind.
25344 (s-selftest-c++): New.
25345 (selftest-c++-gdb): New.
25346 (selftest-c++-valgrind): New.
25348 2017-07-06 Olivier Hainque <hainque@adacore.com>
25350 * gcc.c (process_command): When deciding if undefined variables
25351 should be ignored when processing specs, accept "gcc -v" as well.
25353 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25355 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
25356 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
25358 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25360 * config/arm/arm-cpus.in (armv8-r): Add new entry.
25361 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
25362 * config/arm/arm-tables.opt: Regenerate.
25363 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
25365 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
25367 2017-07-06 Carl Love <cel@us.ibm.com>
25369 * ChangeLog: Clean up from mid air collision
25371 2017-07-06 Carl Love <cel@us.ibm.com>
25373 * config/rs6000/rs6000-c.c: Add support for built-in functions
25374 vector signed int vec_subc (vector signed int, vector signed int);
25375 vector signed __int128 vec_subc (vector signed __int128,
25376 vector signed __int128);
25377 vector unsigned __int128 vec_subc (vector unsigned __int128,
25378 vector unsigned __int128);
25379 vector signed int vec_sube (vector signed int, vector signed int,
25380 vector signed int);
25381 vector unsigned int vec_sube (vector unsigned int,
25382 vector unsigned int,
25383 vector unsigned int);
25384 vector signed __int128 vec_sube (vector signed __int128,
25385 vector signed __int128,
25386 vector signed__int128);
25387 vector unsigned __int128 vec_sube (vector unsigned __int128,
25388 vector unsigned __int128,
25389 vector unsigned __int128);
25390 vector signed int vec_subec (vector signed int, vector signed int,
25391 vector signed int);
25392 vector unsigned int vec_subec (vector unsigned int,
25393 vector unsigned int,
25394 vector unsigned int);
25395 vector signed __int128 vec_subec (vector signed __int128,
25396 vector signed __int128,
25397 vector signed__int128);
25398 vector unsigned __int128 vec_subec (vector unsigned __int128,
25399 vector unsigned __int128,
25400 vector unsigned __int128);
25401 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
25402 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
25403 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
25404 BU_ALTIVEC_OVERLOAD_X definitions.
25405 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
25406 * doc/extend.texi: Update the built-in documentation file for the new
25407 built-in functions.
25409 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25412 * diagnostic-show-locus.c (layout::layout): Use start and finish
25413 spelling location for the start and finish of each range.
25414 * genmatch.c (linemap_client_expand_location_to_spelling_point):
25415 Add unused aspect param.
25416 * input.c (expand_location_1): Add "aspect" param, and use it
25417 to access the correct part of the location.
25418 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
25420 (expand_location_to_spelling_point): Likewise.
25421 (linemap_client_expand_location_to_spelling_point): Add "aspect"
25422 param, and pass it to expand_location_1.
25424 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
25426 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
25427 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
25428 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
25429 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
25430 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
25431 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
25432 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
25433 _mm_maskz_getmant_ss): New intrinsics.
25434 (__builtin_ia32_getexpss128_mask): Changed to ...
25435 __builtin_ia32_getexpss128_round ... this.
25436 (__builtin_ia32_getexpsd128_mask): Changed to ...
25437 __builtin_ia32_getexpsd128_round ... this.
25438 * config/i386/i386-builtin-types.def
25439 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
25440 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
25441 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
25442 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
25443 __builtin_ia32_getmantss_mask_round): New builtins.
25444 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
25445 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
25446 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
25447 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
25448 * config/i386/sse.md
25449 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
25450 avx512f_sgetexp<mode><mask_scalar_name>
25451 <round_saeonly_scalar_name> ... this.
25452 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
25453 %0, %1, %2<round_saeonly_op3>}): Changed to ...
25454 vgetexp<ssescalarmodesuffix>
25455 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25456 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
25457 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
25458 avx512f_vgetmant<mode><mask_scalar_name>
25459 <round_saeonly_scalar_name> ... this.
25460 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
25461 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
25462 vgetmant<ssescalarmodesuffix>
25463 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
25464 %0<mask_scalar_operand4>, %1, %2
25465 <round_saeonly_scalar_mask_op4>, %3} ... this.
25466 * config/i386/subst.md (mask_scalar_operand4,
25467 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
25468 round_saeonly_scalar_nimm_predicate): New subst attributes.
25470 2017-07-06 Julia Koval <julia.koval@intel.com>
25472 * config/i386/i386.c (ix86_erase_embedded_rounding):
25473 Remove code for old rounding pattern.
25475 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
25477 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
25479 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
25481 * doc/sourcebuild.texi (Test Directives, Variants of
25482 dg-require-support): Add documentation for dg-require-stack-check.
25484 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
25486 * config/i386/subst.md (mask_scalar, round_scalar,
25487 round_saeonly_scalar): New meta-templates.
25488 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
25489 round_scalar_mask_operand3, round_scalar_mask_op3,
25490 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
25491 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
25492 round_saeonly_scalar_constraint,
25493 round_saeonly_scalar_prefix): New subst attribute.
25494 * config/i386/sse.md
25495 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
25496 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
25497 <round_scalar_name> ... this.
25498 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
25499 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
25500 <round_scalar_name> ... this.
25501 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
25502 <sse>_vm<code><mode>3<mask_scalar_name>
25503 <round_saeonly_scalar_name> ... this.
25504 (v<plusminus_mnemonic><ssescalarmodesuffix>
25505 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25506 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25507 v<plusminus_mnemonic><ssescalarmodesuffix>
25508 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25509 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25510 (v<multdiv_mnemonic><ssescalarmodesuffix>
25511 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25512 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25513 v<multdiv_mnemonic><ssescalarmodesuffix>
25514 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25515 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25516 (v<maxmin_float><ssescalarmodesuffix>
25517 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
25518 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
25519 v<maxmin_float><ssescalarmodesuffix>
25520 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25521 %0<mask_scalar_operand3>, %1, %<iptr>2
25522 <round_saeonly_scalar_mask_op3>} ... this.
25524 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
25526 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
25527 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
25529 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
25530 Alan Hayward <alan.hayward@arm.com>
25531 David Sherwood <david.sherwood@arm.com>
25533 * combine.c (simplify_if_then_else): Remove "enum" before
25535 * compare-elim.c (can_eliminate_compare): Likewise.
25536 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
25538 (aarch64_lookup_simd_builtin_type): Likewise.
25539 (aarch64_simd_builtin_type): Likewise.
25540 (aarch64_init_simd_builtin_types): Likewise.
25541 (aarch64_simd_expand_args): Likewise.
25542 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
25544 (aarch64_reverse_mask): Likewise.
25545 (aarch64_simd_emit_reg_reg_move): Likewise.
25546 (aarch64_gen_adjusted_ldpstp): Likewise.
25547 (aarch64_ccmp_mode_to_code): Likewise.
25548 (aarch64_operands_ok_for_ldpstp): Likewise.
25549 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25550 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
25552 (aarch64_min_divisions_for_recip_mul): Likewise.
25553 (aarch64_reassociation_width): Likewise.
25554 (aarch64_get_condition_code_1): Likewise.
25555 (aarch64_simd_emit_reg_reg_move): Likewise.
25556 (aarch64_simd_attr_length_rglist): Likewise.
25557 (aarch64_reverse_mask): Likewise.
25558 (aarch64_operands_ok_for_ldpstp): Likewise.
25559 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25560 (aarch64_gen_adjusted_ldpstp): Likewise.
25561 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
25563 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
25564 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
25565 (arm_lookup_simd_builtin_type): Likewise.
25566 (arm_simd_builtin_type): Likewise.
25567 (arm_init_simd_builtin_types): Likewise.
25568 (arm_expand_builtin_args): Likewise.
25569 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
25570 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
25571 (ft32_setup_incoming_varargs): Likewise.
25572 (ft32_function_arg): Likewise.
25573 (ft32_function_arg_advance): Likewise.
25574 (ft32_pass_by_reference): Likewise.
25575 (ft32_arg_partial_bytes): Likewise.
25576 (ft32_valid_pointer_mode): Likewise.
25577 (ft32_addr_space_pointer_mode): Likewise.
25578 (ft32_addr_space_legitimate_address_p): Likewise.
25579 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
25581 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
25582 (ix86_emit_outlined_ms2sysv_restore): Likewise.
25583 (iamcu_alignment): Likewise.
25584 (canonicalize_vector_int_perm): Likewise.
25585 (ix86_noce_conversion_profitable_p): Likewise.
25586 (ix86_mpx_bound_mode): Likewise.
25587 (ix86_operands_ok_for_move_multiple): Likewise.
25588 * config/microblaze/microblaze-protos.h
25589 (microblaze_expand_conditional_branch_reg): Likewise.
25590 * config/microblaze/microblaze.c
25591 (microblaze_expand_conditional_branch_reg): Likewise.
25592 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
25594 (rs6000_reassociation_width): Likewise.
25595 (rs6000_invalid_binary_op): Likewise.
25596 (fusion_p9_p): Likewise.
25597 (emit_fusion_p9_load): Likewise.
25598 (emit_fusion_p9_store): Likewise.
25599 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
25601 (riscv_hard_regno_mode_ok_p): Likewise.
25602 (riscv_address_insns): Likewise.
25603 (riscv_split_symbol): Likewise.
25604 (riscv_legitimize_move): Likewise.
25605 (riscv_function_value): Likewise.
25606 (riscv_hard_regno_nregs): Likewise.
25607 (riscv_expand_builtin): Likewise.
25608 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
25609 (riscv_build_integer): Likewise.
25610 (riscv_split_integer): Likewise.
25611 (riscv_legitimate_constant_p): Likewise.
25612 (riscv_cannot_force_const_mem): Likewise.
25613 (riscv_regno_mode_ok_for_base_p): Likewise.
25614 (riscv_valid_base_register_p): Likewise.
25615 (riscv_valid_offset_p): Likewise.
25616 (riscv_valid_lo_sum_p): Likewise.
25617 (riscv_classify_address): Likewise.
25618 (riscv_legitimate_address_p): Likewise.
25619 (riscv_address_insns): Likewise.
25620 (riscv_load_store_insns): Likewise.
25621 (riscv_force_binary): Likewise.
25622 (riscv_split_symbol): Likewise.
25623 (riscv_force_address): Likewise.
25624 (riscv_legitimize_address): Likewise.
25625 (riscv_move_integer): Likewise.
25626 (riscv_legitimize_const_move): Likewise.
25627 (riscv_legitimize_move): Likewise.
25628 (riscv_address_cost): Likewise.
25629 (riscv_subword): Likewise.
25630 (riscv_output_move): Likewise.
25631 (riscv_canonicalize_int_order_test): Likewise.
25632 (riscv_emit_int_order_test): Likewise.
25633 (riscv_function_arg_boundary): Likewise.
25634 (riscv_pass_mode_in_fpr_p): Likewise.
25635 (riscv_pass_fpr_single): Likewise.
25636 (riscv_pass_fpr_pair): Likewise.
25637 (riscv_get_arg_info): Likewise.
25638 (riscv_function_arg): Likewise.
25639 (riscv_function_arg_advance): Likewise.
25640 (riscv_arg_partial_bytes): Likewise.
25641 (riscv_function_value): Likewise.
25642 (riscv_pass_by_reference): Likewise.
25643 (riscv_setup_incoming_varargs): Likewise.
25644 (riscv_print_operand): Likewise.
25645 (riscv_elf_select_rtx_section): Likewise.
25646 (riscv_save_restore_reg): Likewise.
25647 (riscv_for_each_saved_reg): Likewise.
25648 (riscv_register_move_cost): Likewise.
25649 (riscv_hard_regno_mode_ok_p): Likewise.
25650 (riscv_hard_regno_nregs): Likewise.
25651 (riscv_class_max_nregs): Likewise.
25652 (riscv_memory_move_cost): Likewise.
25653 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
25654 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
25655 (rl78_addr_space_address_mode): Likewise.
25656 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
25658 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
25659 (rs6000_reassociation_width): Likewise.
25660 (rs6000_invalid_binary_op): Likewise.
25661 (fusion_p9_p): Likewise.
25662 (emit_fusion_p9_load): Likewise.
25663 (emit_fusion_p9_store): Likewise.
25664 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
25665 (ok_for_simple_move_operands): Likewise.
25666 (ok_for_simple_move_strict_operands): Likewise.
25667 (ok_for_simple_arith_logic_operands): Likewise.
25668 (visium_legitimize_reload_address): Likewise.
25669 (visium_select_cc_mode): Likewise.
25670 (output_cbranch): Likewise.
25671 (visium_split_double_move): Likewise.
25672 (visium_expand_copysign): Likewise.
25673 (visium_expand_int_cstore): Likewise.
25674 (visium_expand_fp_cstore): Likewise.
25675 * config/visium/visium.c (visium_pass_by_reference): Likewise.
25676 (visium_function_arg): Likewise.
25677 (visium_function_arg_advance): Likewise.
25678 (visium_libcall_value): Likewise.
25679 (visium_setup_incoming_varargs): Likewise.
25680 (visium_legitimate_constant_p): Likewise.
25681 (visium_legitimate_address_p): Likewise.
25682 (visium_legitimize_address): Likewise.
25683 (visium_secondary_reload): Likewise.
25684 (visium_register_move_cost): Likewise.
25685 (visium_memory_move_cost): Likewise.
25686 (prepare_move_operands): Likewise.
25687 (ok_for_simple_move_operands): Likewise.
25688 (ok_for_simple_move_strict_operands): Likewise.
25689 (ok_for_simple_arith_logic_operands): Likewise.
25690 (visium_function_value_1): Likewise.
25691 (rtx_ok_for_offset_p): Likewise.
25692 (visium_legitimize_reload_address): Likewise.
25693 (visium_split_double_move): Likewise.
25694 (visium_expand_copysign): Likewise.
25695 (visium_expand_int_cstore): Likewise.
25696 (visium_expand_fp_cstore): Likewise.
25697 (visium_split_cstore): Likewise.
25698 (visium_select_cc_mode): Likewise.
25699 (visium_split_cbranch): Likewise.
25700 (output_cbranch): Likewise.
25701 (visium_print_operand_address): Likewise.
25702 * expmed.c (flip_storage_order): Likewise.
25703 * expmed.h (emit_cstore): Likewise.
25704 (flip_storage_order): Likewise.
25705 * genrecog.c (validate_pattern): Likewise.
25706 * hsa-gen.c (gen_hsa_addr): Likewise.
25707 * internal-fn.c (expand_arith_overflow): Likewise.
25708 * ira-color.c (allocno_copy_cost_saving): Likewise.
25709 * lra-assigns.c (find_hard_regno_for_1): Likewise.
25710 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
25711 (process_invariant_for_inheritance): Likewise.
25712 * lra-eliminations.c (move_plus_up): Likewise.
25713 * omp-low.c (lower_oacc_reductions): Likewise.
25714 * simplify-rtx.c (simplify_subreg): Likewise.
25715 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
25716 (TARGET_CHKP_BOUND_MODE): Likewise..
25717 * targhooks.c (default_chkp_bound_mode): Likewise.
25718 (default_setup_incoming_vararg_bounds): Likewise.
25719 * targhooks.h (default_chkp_bound_mode): Likewise.
25720 (default_setup_incoming_vararg_bounds): Likewise.
25721 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
25722 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25723 (have_whole_vector_shift): Likewise.
25724 * tree-vect-stmts.c (vectorizable_load): Likewise.
25725 * doc/tm.texi: Regenerate.
25727 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25729 Graceful degrade if Binutils PR21472 is not available.
25732 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
25733 .rodata in flash test fails.
25734 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
25735 * confgure: Regenerate.
25736 * config.in: Regenerate.
25737 * config/avr/avr.c (avr_asm_named_section)
25738 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
25739 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
25740 (avr_asm_init_sections): Same.
25742 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25744 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
25745 (fma<VH:mode>4_intrinsic): Likewise.
25746 (*fmsub<VCVTF:mode>4): Likewise.
25747 (*fmsub<VH:mode>4_intrinsic): Likewise.
25749 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25752 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
25753 Don't depend on "optimize > 0".
25754 (out_movhi_r_mr, out_movqi_mr_r): Same.
25755 (out_movhi_mr_r, out_movqi_r_mr): Same.
25756 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
25757 io_address_operand on "optimize > 0".
25759 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25761 * tree-loop-distribution.c: Add general explanantion on the pass.
25762 (generate_loops_for_partition): Mark distributed loop.
25763 (pg_add_dependence_edges): New parameter. Handle alias data
25764 dependence specially and record it in the parameter if asked.
25765 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
25766 (init_partition_graph_vertices, add_partition_graph_edge): New.
25767 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
25768 (free_partition_graph_vdata, build_partition_graph): New.
25769 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
25770 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
25771 (data_ref_segment_size, latch_dominated_by_data_ref): New.
25772 (compute_alias_check_pairs, version_loop_by_alias_check): New.
25773 (version_for_distribution_p, finalize_partitions): New.
25774 (distribute_loop): Handle alias data dependence specially. Factor
25775 out loop fusion code as functions and call these functions.
25777 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25779 * tree-loop-distribution.c (classify_partition): New parameter and
25780 better handle reduction statement.
25781 (rdg_build_partitions): Revise comment.
25782 (distribute_loop): Compute statements in all partitions and pass it
25783 to classify_partition.
25785 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25787 * tree-loop-distribution.c (enum partition_type): New.
25788 (struct partition): New field type.
25789 (partition_merge_into): Add parameter. Update partition type.
25790 (data_dep_in_cycle_p, update_type_for_merge): New functions.
25791 (build_rdg_partition_for_vertex): Compute partition type.
25792 (rdg_build_partitions): Dump partition type.
25793 (distribute_loop): Update calls to partition_merge_into.
25795 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25797 * tree-loop-distribution.c (struct ddr_hasher): New.
25798 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
25800 (classify_partition): Call get_data_dependence.
25801 (pg_add_dependence_edges): Ditto.
25802 (distribute_loop): Release data dependence hash table.
25804 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25806 * tree-loop-distribution.c (ref_base_address): Delete.
25807 (similar_memory_accesses): Rename ...
25808 (share_memory_accesses): ... to this. Check if partitions access
25809 the same memory reference.
25810 (distribute_loop): Call share_memory_accesses.
25812 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25814 * tree-loop-distribution.c (struct partition): New field recording
25815 its data reference.
25816 (partition_alloc, partition_free): Init and release data refs.
25817 (partition_merge_into): Merge data refs.
25818 (build_rdg_partition_for_vertex): Collect data refs for partition.
25819 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25821 (distribute_loop): Remve data refs from vertice data of partition
25824 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25826 * tree-loop-distribution.c (params.h): Include header file.
25827 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25828 (datarefs_vec): New global var.
25829 (create_rdg_vertices): Use datarefs_vec directly.
25830 (free_rdg): Don't free data references.
25831 (build_rdg): Update use. Don't free data references.
25832 (distribute_loop): Compute global variable for data references.
25833 Bail out if there are too many data references.
25835 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25837 * tree-loop-distribution.c (loop_nest): New global var.
25838 (build_rdg): Use loop directly, rather than loop nest.
25839 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25841 (distribute_loop): Compute global variable loop nest. Update use.
25843 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25845 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
25846 (partition_merge_into): New parameter. Dump reason for fusion.
25847 (distribute_loop): Update use of partition_merge_into.
25849 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25851 * tree-loop-distribution.c (bb_top_order_index): New.
25852 (bb_top_order_index_size, bb_top_order_cmp): New.
25853 (stmts_from_loop): Use topological order.
25854 (pass_loop_distribution::execute): Compute and release topological
25855 order for basic blocks.
25857 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25859 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25862 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25864 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25865 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25866 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25867 * internal-fn.def (LOOP_DIST_ALIAS): New.
25868 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25869 (fold_loop_internal_call): ... this.
25870 (vect_loop_dist_alias_call): New function.
25871 (set_uid_loop_bbs): Call fold_loop_internal_call.
25872 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25875 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25878 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25879 Require dead FLAGS_REG at the beginning of a peephole.
25881 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25884 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25885 arguments in the call to __builtin_ia32_sbb_u32.
25886 (_subborrow_u64): Swap _X and _Y arguments in the call to
25887 __builtin_ia32_sbb_u64.
25889 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25892 * tree-vrp.c (compare_assert_loc): Turn into a function template
25893 with stable template parameter. Only test if a->e is NULL,
25894 !a->e == !b->e has been verified already. Use e == NULL or
25895 e != NULL instead of e or ! e tests. If stable is true, don't use
25896 iterative_hash_expr, on the other side allow a or b or both NULL
25897 and sort the NULLs last.
25898 (process_assert_insertions): Sort using compare_assert_loc<false>
25899 instead of compare_assert_loc, later sort using
25900 compare_assert_loc<true> before calling process_assert_insertions_for
25901 in a loop. Use break instead of continue once seen NULL pointer.
25903 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25905 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25906 Cortex-R7 and Cortex-R8 processors.
25908 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25910 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25911 uninitialized while src is not.
25913 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
25915 * common/config/arm/arm-common.c: Adjust include path for
25917 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
25918 (arm-cpu.h): Create in build directory. Adjust dependency rules.
25919 (arm-cpu-data.h): Likewise.
25920 (arm-cpu-cdata.h): Likewise.
25921 * config/arm/arm-cpu.h: Delete.
25922 * config/arm/arm-cpu-cdata.h: Delete.
25923 * config/arm/arm-cpu-data.h: Delete.
25925 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
25927 * config/arm/arm-cpus.in (cortex-a55): New.
25928 (cortex-a75): Likewise.
25929 (cortex-a75.cortex-a55): Likewise.
25930 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
25932 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
25933 * config/arm/arm-cpu-cdata.h: Regenerate.
25934 * config/arm/arm-cpu-data.h: Regenerate.
25935 * config/arm/arm-cpu.h: Regenerate.
25936 * config/arm/arm-tables.opt: Regenerate.
25937 * config/arm/arm-tune.md: Regenerate.
25939 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25941 * haifa-sched.c (sched_create_recovery_edges): Update profile.
25943 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25945 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
25948 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
25950 PR tree-optimization/81292
25951 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
25952 full_string_p, also call adjust_related_strinfos if the adjustment
25953 is simple, otherwise invalidate related strinfos.
25955 2017-07-04 Martin Liska <mliska@suse.cz>
25958 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
25959 newly created variable as DECL_IGNORED_P.
25961 2017-07-04 Martin Liska <mliska@suse.cz>
25964 * ipa-inline.c (inline_small_functions):
25965 Use xstrdup_for_dump.
25967 2017-07-04 Tom de Vries <tom@codesourcery.com>
25969 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
25971 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25974 * config/darwin.c (darwin_function_switched_text_sections):
25977 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
25979 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
25981 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25983 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
25985 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25987 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
25988 min_profitable_iters, and th as inclusive lower bounds.
25989 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
25990 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
25991 for min_profitable_iters and min_profitable_estimate.
25992 (vect_transform_loop): Treat th as an inclusive lower bound.
25993 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
25995 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25998 * config/darwin.c (darwin_function_switched_text_sections):
25999 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
26000 in two pieces, and suppress the use of buf.
26002 2017-07-03 Nathan Sidwell <nathan@acm.org>
26004 * hash-table.h (hash_table_mod1): Fix indentation.
26006 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26008 PR middle-end/81290
26009 * predict.c (force_edge_cold): Be more careful about propagation
26011 * profile-count.h (profile_probability::guessed,
26012 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
26014 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
26016 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
26018 * doc/invoke.texi (rcpc architecture extension): Document it.
26020 2017-07-03 Richard Biener <rguenther@suse.de>
26022 PR tree-optimization/60510
26023 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
26024 the scalar reduction PHI and use it.
26025 (vectorizable_reduction): Properly guard the single_defuse_cycle
26026 path for non-SLP reduction chains where we cannot use it.
26027 Rework reduc_def/index and vector type deduction. Rework
26028 vector operand gathering during reduction op code-gen.
26029 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
26030 chains dissolve the chain and leave it to non-SLP reduction
26033 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26035 * tree-data-ref.h (dr_alignment): Declare.
26036 * tree-data-ref.c (dr_alignment): New function.
26037 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
26038 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
26040 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
26042 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26044 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
26045 and base_misalignment fields.
26046 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
26047 * tree-data-ref.c: Include builtins.h.
26048 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
26049 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
26050 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
26051 * tree-vect-data-refs.c: Include tree-cfg.h.
26052 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
26053 fields instead of calculating an alignment here.
26054 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
26055 innermost_loop_behavior fields.
26057 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26059 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
26061 (DR_STEP_ALIGNMENT): New macro.
26062 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
26063 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
26064 (create_data_ref): Print it.
26065 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
26066 to tell whether the step preserves vector (mis)alignment.
26067 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
26068 Move the check for an integer step and generalise to all INTEGER_CST.
26069 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
26070 Print the outer step alignment.
26072 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26074 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
26075 with offset_alignment.
26076 (DR_ALIGNED_TO): Delete.
26077 (DR_OFFSET_ALIGNMENT): New macro.
26078 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
26079 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
26080 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
26081 (create_data_ref): Likewise.
26082 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
26083 (vect_analyze_data_refs): Likewise.
26084 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
26085 creating dummy innermost behavior.
26087 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26089 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
26090 with a "innermost_loop_behavior *" and refeence tree.
26091 * tree-data-ref.c (dr_analyze_innermost): Likewise.
26092 (create_data_ref): Update call accordingly.
26093 * tree-predcom.c (find_looparound_phi): Likewise.
26095 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26097 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
26098 fields with dr_wrt_vec_loop.
26099 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
26100 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
26101 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
26102 (vect_dr_behavior): New function.
26103 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26104 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
26105 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
26106 track whether the step preserves the misalignment.
26107 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26108 Use vect_dr_behavior.
26109 (vect_setup_realignment): Update call accordingly.
26110 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
26111 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
26112 call to vect_create_addr_base_for_vector_ref.
26113 (vect_create_cond_for_align_checks): Likewise.
26114 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
26115 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
26116 (vect_recog_mask_conversion_pattern): Likewise.
26117 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
26118 (new_stmt_vec_info): Remove redundant zeroing.
26120 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
26122 * common/config/arm/arm-common.c (arm_be8_option): New function.
26123 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
26124 (ISA_ARMv6): Add isa_bit_be8.
26125 * config/arm/arm.h (arm_be8_option): Add prototype.
26126 (BE8_SPEC_FUNCTION): New define.
26127 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
26128 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
26129 (mlittle-endian): Similarly.
26130 (mbe8, mbe32): New options.
26131 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
26132 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
26134 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26136 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
26138 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26140 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
26141 (cleanup_tree_cfg_bb): Use it.
26142 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
26144 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
26146 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26149 * loop-doloop.c (add_test): Update profile.
26151 2017-07-03 Martin Liska <mliska@suse.cz>
26154 * sanopt.c (rewrite_usage_of_param): New function.
26155 (sanitize_rewrite_addressable_params): Likewise.
26156 (pass_sanopt::execute): Call rewrite_usage_of_param.
26158 2017-07-03 Richard Biener <rguenther@suse.de>
26160 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
26161 back to using VIEW_CONVERT_EXPR.
26163 2017-07-03 Martin Liska <mliska@suse.cz>
26166 * doc/extend.texi: Document when a resolver function is
26167 generated for target_clones.
26169 2017-07-03 Martin Liska <mliska@suse.cz>
26171 * asan.c (asan_emit_stack_protection): Unpoison just red zones
26172 and shadow memory of auto variables which are subject of
26173 use-after-scope sanitization.
26174 (asan_expand_mark_ifn): Add do set only when is_poison.
26176 2016-07-03 Richard Biener <rguenther@suse.de>
26178 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
26180 (vect_force_simple_reduction): Record reduction def -> phi mapping.
26181 (vectorizable_reduction): Perform reduction PHI creation when
26182 visiting a reduction PHI and adjust and simplify code generation
26183 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
26184 (vect_transform_loop): Visit reduction PHIs.
26185 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
26186 defs into the SLP tree.
26187 (vect_build_slp_tree): Reduction defs terminate the recursion.
26188 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
26190 (vect_get_vec_defs_for_stmt_copy): Export.
26191 (vect_get_vec_defs): Likewise.
26192 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
26194 (vect_get_vec_defs_for_stmt_copy): Declare.
26195 (vect_get_vec_defs): Likewise.
26197 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26199 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
26200 parameter with a "loop" parameter and use it instead of the
26201 loop containing DR_STMT. Don't check simple_iv when doing
26202 BB analysis. Describe the two analysis modes in the comment.
26204 2017-07-03 Tom de Vries <tom@codesourcery.com>
26206 PR tree-optimization/69468
26207 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
26208 (find_same_succ_bb): Handle ignore_edge_flags.
26210 2017-07-03 Tom de Vries <tom@codesourcery.com>
26212 PR tree-optimization/81192
26213 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
26215 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
26217 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
26219 2017-07-03 Tom de Vries <tom@codesourcery.com>
26221 PR tree-optimization/81192
26222 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
26223 BB_SAME_SUCC (bb) == NULL.
26225 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26227 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
26230 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26232 * dumpfile.c: Include profile-count.h
26233 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
26235 (insert_cond_bb): Update profile.
26236 * tree-cfg.h (insert_cond_bb): Update prototype.
26237 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
26238 * tree-dump.c: Do not include tree-cfg.
26240 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26242 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
26244 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26246 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
26249 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26251 * tree-complex.c (expand_complex_div_wide): update profile.
26253 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26254 Alan Hayward <alan.hayward@arm.com>
26255 David Sherwood <david.sherwood@arm.com>
26257 * Makefile.in (MACHMODE_H): Remove insn-modes.h
26258 (CORETYPES_H): New define.
26259 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
26260 (insn-modes-inline.h, s-modes-inline-h): New rules.
26261 (generated_files): Add insn-modes-inline.h.
26262 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
26263 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
26264 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
26265 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
26266 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
26267 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
26268 (build/gencodes.o, build/genconditions.o): Likewise.
26269 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
26270 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
26271 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
26272 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
26273 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
26274 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
26275 * coretypes.h: Include everything up to real.h for generators.
26276 Include insn-modes.h first. Include wide-int-print.h after
26277 wide-int.h. Include insn-modes-inline.h and then machmode.h.
26278 * machmode.h: Don't include insn-modes.h here.
26279 * function-tests.c: Remove includes of signop.h, machmode.h,
26280 double-int.h and wide-int.h.
26282 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
26284 * optc-save-gen.awk: Likewise.
26285 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
26286 * godump.c: Remove include of wide-int-print.h.
26287 * pretty-print.h: Likewise.
26288 * wide-int-print.cc: Likewise.
26289 * wide-int.cc: Likewise.
26290 * hash-map-tests.c: Remove include of signop.h.
26291 * hash-set-tests.c: Likewise.
26292 * rtl-tests.c: Likewise.
26293 * mkconfig.sh: Remove include of machmode.h.
26294 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
26296 (emit_insn_modes_inline_h): ...this new function. Emit the code
26297 into an insn-modes-inline.h header file, adding appropriate
26298 include guards and end comments.
26299 (emit_insn_modes_c_header): Remove include of machmode.h.
26300 (emit_min_insn_modes_c_header): Include coretypes.h rather than
26302 (main): Handle -i flag and call emit_insn_modes_inline_h when
26305 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26307 * tree-ssa-strlen.c (strinfo): Rename the length field to
26308 nonzero_chars. Add a full_string_p field.
26309 (compare_nonzero_chars, zero_length_string_p): New functions.
26310 (get_addr_stridx): Add an offset_out parameter.
26311 Use compare_nonzero_chars.
26312 (get_stridx): Update accordingly. Use compare_nonzero_chars.
26313 (new_strinfo): Update after above changes to strinfo.
26314 (set_endptr_and_length): Set full_string_p.
26315 (get_string_length): Update after above changes to strinfo.
26316 (unshare_strinfo): Update call to new_strinfo.
26317 (maybe_invalidate): Likewise.
26318 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
26319 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
26320 as a uhwi instead of an shwi. Update after above changes to
26321 strinfo and new_strinfo.
26322 (zero_length_string): Assert that chainsi contains full strings.
26323 Use zero_length_string_p. Update call to new_strinfo.
26324 (adjust_related_strinfos): Update after above changes to strinfo.
26325 Copy full_string_p from origsi.
26326 (adjust_last_stmt): Use zero_length_string_p.
26327 (handle_builtin_strlen): Update after above changes to strinfo and
26328 new_strinfo. Install the lhs as the string length if the previous
26329 entry didn't describe a full string.
26330 (handle_builtin_strchr): Update after above changes to strinfo
26332 (handle_builtin_strcpy): Likewise.
26333 (handle_builtin_strcat): Likewise.
26334 (handle_builtin_malloc): Likewise.
26335 (handle_pointer_plus): Likewise.
26336 (handle_builtin_memcpy): Likewise. Track nonzero characters
26337 that aren't necessarily followed by a nul terminator.
26338 (handle_char_store): Likewise.
26340 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26342 PR tree-optimization/80769
26343 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
26344 for malloc and calloc. Document the new invariant that all related
26345 strinfos have delayed lengths or none do.
26346 (verify_related_strinfos): Move earlier in file.
26347 (set_endptr_and_length): New function, split out from...
26348 (get_string_length): ...here. Also set the lengths of related
26350 (zero_length_string): Assert that chainsi has known (rather than
26352 (adjust_related_strinfos): Likewise.
26354 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26356 PR tree-optimization/81136
26357 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
26358 assert that two references with the same misalignment have the same
26359 compile-time misalignment if those compile-time misalignments
26362 2017-07-01 Andi Kleen <ak@linux.intel.com>
26364 * print-tree.c (print_node): Print all attributes.
26366 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26368 * cfg.c (scale_bbs_frequencies): New function.
26369 * cfg.h (scale_bbs_frequencies): Declare it.
26370 * cfgloopanal.c (single_likely_exit): Cleanup.
26371 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
26373 (scale_loop_profile): Likewise.
26374 (loop_version): Likewise.
26375 (create_empty_loop_on_edge): Update.
26376 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
26377 scale_loop_frequencies, scale_loop_profile, loopify,
26378 loop_version): Update prototypes.
26379 * modulo-sched.c (sms_schedule): Update.
26380 * predict.c (unlikely_executed_edge_p): Also check probability.
26381 (probably_never_executed_edge_p): Fix typo.
26382 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26383 * tree-parloops.c (gen_parallel_loop): Update.
26384 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
26385 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26386 * tree-ssa-loop-split.c (split_loop): Update.
26387 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26388 * tree-vect-loop-manip.c (vect_do_peeling): Update.
26389 (vect_loop_versioning): Update.
26390 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26392 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26394 * trans-mem.c (split_bb_make_tm_edge): Update profile.
26396 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26398 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
26399 to keep profile consistent.
26401 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26403 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
26404 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
26405 * profile-count.h (max_safe_multiplier): Make unsigned.
26406 (profile_count::guessed_zero): New.
26408 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26410 * bb-reorder.c (fix_up_crossing_landing_pad,
26411 fix_crossing_conditional_branches): Use make_single_succ_edge
26412 to keep profile consistent.
26414 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26416 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
26419 2017-07-01 Jakub Jelinek <jakub@redhat.com>
26422 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
26423 the right scopes, make sure cond_jump isn't preserved between multiple
26424 iterations. Search for fallthru edge whenever there are 3+ edges and
26425 use find_fallthru_edge for it.
26427 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26429 Patch by Alexander Monakov <amonakov@ispras.ru>
26430 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
26431 probabilities consistently.
26433 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26435 * pa.c (pa_expand_compare_and_swap_loop): Update call of
26436 emit_cmp_and_jump_insns.
26438 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26441 * tree-inline.c (expand_call_inline): Combine profile statuses.
26443 2017-06-30 Andrew Pinski <apinski@cavium.com>
26445 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
26446 fold_stmt returned true.
26448 2017-06-30 Nathan Sidwell <nathan@acm.org>
26450 * ggc.h (empty_string): Delete.
26451 * cfgexpand.c (expand_asm_stmt): Use plain "".
26452 * optabs.c (expand_asm_memory_barrier): Likewise.
26453 * stringpool.c (empty_string): Delete.
26454 (digit_vector, digit_string): Delete.
26455 (ggc_alloc_string): Use plain "", don't optimize single digit
26456 strings. Use ggc_alloc_atomic.
26458 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
26460 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
26461 comparison set and one other set, use the cost of the non-comparison
26464 2017-06-30 Nathan Sidwell <nathan@acm.org>
26466 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
26469 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
26471 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
26472 loops. Remove now unneeded calls to gimple_switch_set_label() that
26473 just set removed labels to NULL_TREE.
26475 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
26477 * tree-ssanames.c (set_range_info_raw): Abstract from ...
26478 (set_range_info): ...here. Only call set_range_info_raw if domain
26480 (set_nonzero_bits): Call set_range_info_raw.
26481 * tree-ssanames.h (set_range_info_raw): New.
26483 2017-06-30 Jakub Jelinek <jakub@redhat.com>
26486 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
26487 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
26488 of nonimmediate_operand and <store_mask_constraint> instead of m
26489 for the input operand. For V8FI iterator, always split if input
26490 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
26491 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
26492 <store_mask_predicate> instead of register_operand and
26493 <store_mask_constraint> instead of v for the input operand. Make
26494 sure both operands aren't MEMs for if not <mask_applied>.
26496 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
26498 * lto-wrapper.c (copy_file) Close both file descriptors before
26501 2017-06-30 Martin Liska <mliska@suse.cz>
26504 * multiple_target.c (create_dispatcher_calls): Make ifunc
26505 also for function that don't have calls or are not referenced.
26507 2017-06-30 Richard Biener <rguenther@suse.de>
26509 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
26510 analyze the first scalar stmt. Move vector type computation
26511 for the BB case here from ...
26512 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
26513 live operation processing in the SLP case properly.
26515 2017-06-30 Richard Biener <rguenther@suse.de>
26517 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
26519 2017-06-30 Martin Liska <mliska@suse.cz>
26522 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
26523 before BUILT_IN_UNWIND_RESUME when ASAN is used.
26525 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
26527 * doc/invoke.texi (AArch64): Add missing options and remove redundant
26530 2017-06-30 Richard Biener <rguenther@suse.de>
26532 PR tree-optimization/81249
26533 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
26534 condition reduction result to original scalar type.
26536 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26538 * profile-count.h (enum profile_quality): Fix typos and whitespace
26541 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26543 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
26544 type for branch probabilities.
26546 2017-06-29 Julian Brown <julian@codesourcery.com>
26547 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26549 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
26550 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
26551 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
26552 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
26554 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26556 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
26557 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
26558 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
26559 CC usage from generic code to here.
26560 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
26561 CC usage into the target macros.
26563 2017-06-29 Maya Rashish <coypu@sdf.org>
26565 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
26568 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26570 * arm/arm-builtins.c: Include profile-count.h
26571 * except.c (sjlj_emit_function_enter): Use
26572 profile_probability::unlikely.
26574 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26576 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
26577 and tocrel_offset be pointer args rather than implicitly using
26579 (legitimate_constant_pool_address_p, rs6000_emit_move,
26580 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
26581 tocrel_offset and use in toc_relative_expr_p call.
26582 (print_operand, print_operand_address): Use static tocrel_base_oac
26583 and tocrel_offset_oac.
26584 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
26587 2017-06-29 Maya Rashish <coypu@sdf.org>
26589 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
26591 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
26593 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
26594 objects, take into account only the alignment of 'op0' and 'mode1' if
26597 2017-06-29 Steve Ellcey <sellcey@cavium.com>
26599 * ccmp.c (ccmp_tree_comparison_p): New function.
26600 (ccmp_candidate_p): Update to use above function.
26601 (get_compare_parts): New function.
26602 (expand_ccmp_next): Update to use new functions.
26603 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
26605 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
26606 take mode as argument.
26607 * ccmp.h (expand_ccmp_expr): Add mode as argument.
26608 * expr.c (expand_expr_real_1): Pass mode as argument.
26610 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
26612 * combine.c (combine_instructions): Print insns to dump_file, together
26615 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26617 * asan.c (asan_emit_stack_protection): Update.
26618 (create_cond_insert_point): Update.
26619 * auto-profile.c (afdo_propagate_circuit): Update.
26620 * basic-block.h (struct edge_def): Turn probability to
26621 profile_probability.
26622 (EDGE_FREQUENCY): Update.
26623 * bb-reorder.c (find_traces_1_round): Update.
26624 (better_edge_p): Update.
26625 (sanitize_hot_paths): Update.
26626 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
26627 (make_single_succ_edge): Update.
26628 (check_bb_profile): Update.
26629 (dump_edge_info): Update.
26630 (update_bb_profile_for_threading): Update.
26631 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
26632 probabilitycount to 0.
26633 * cfgbuild.c (compute_outgoing_frequencies): Update.
26634 * cfgcleanup.c (try_forward_edges): Update.
26635 (outgoing_edges_match): Update.
26636 (try_crossjump_to_edge): Update.
26637 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
26638 (expand_gimple_tailcall): Update.
26639 (construct_init_block): Use make_single_succ_edge.
26640 (construct_exit_block): Use make_single_succ_edge.
26641 * cfghooks.c (verify_flow_info): Update.
26642 (redirect_edge_succ_nodup): Update.
26643 (split_edge): Update.
26644 (account_profile_record): Update.
26645 * cfgloopanal.c (single_likely_exit): Update.
26646 * cfgloopmanip.c (scale_loop_profile): Update.
26647 (set_zero_probability): Remove.
26648 (duplicate_loop_to_header_edge): Update.
26649 * cfgloopmanip.h (loop_version): Update prototype.
26650 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
26651 (force_nonfallthru_and_redirect): Update.
26652 (update_br_prob_note): Update.
26653 (rtl_verify_edges): Update.
26654 (purge_dead_edges): Update.
26655 (rtl_lv_add_condition_to_bb): Update.
26656 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
26657 * cgraphunit.c (init_lowered_empty_function): Update.
26658 (cgraph_node::expand_thunk): Update.
26659 * cilk-common.c: Include profile-count.h
26660 * dojump.c (inv): Remove.
26661 (jumpifnot): Update.
26662 (jumpifnot_1): Update.
26663 (do_jump_1): Update.
26665 (do_jump_by_parts_greater_rtx): Update.
26666 (do_compare_rtx_and_jump): Update.
26667 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
26668 do_jump_1. do_compare_rtx_and_jump): Update prototype.
26669 * dwarf2cfi.c: Include profile-count.h
26670 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
26671 (sjlj_emit_dispatch_table): Likewise.
26672 * explow.c: Include profile-count.h
26673 * expmed.c (emit_store_flag_force): Update.
26674 (do_cmp_and_jump): Update.
26675 * expr.c (compare_by_pieces_d::generate): Update.
26676 (compare_by_pieces_d::finish_mode): Update.
26677 (emit_block_move_via_loop): Update.
26678 (store_expr_with_bounds): Update.
26679 (store_constructor): Update.
26680 (expand_expr_real_2): Update.
26681 (expand_expr_real_1): Update.
26682 * expr.h (try_casesi, try_tablejump): Update prototypes.
26683 * gimple-pretty-print.c (dump_probability): Update.
26684 (dump_profile): New.
26685 (dump_gimple_label): Update.
26686 (dump_gimple_bb_header): Update.
26687 * graph.c (draw_cfg_node_succ_edges): Update.
26688 * hsa-gen.c (convert_switch_statements): Update.
26689 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
26690 (find_if_case_1): Update.
26691 (find_if_case_2): Update.
26692 * internal-fn.c (expand_arith_overflow_result_store): Update.
26693 (expand_addsub_overflow): Update.
26694 (expand_neg_overflow): Update.
26695 (expand_mul_overflow): Update.
26696 (expand_vector_ubsan_overflow): Update.
26697 * ipa-cp.c (good_cloning_opportunity_p): Update.
26698 * ipa-split.c (split_function): Use make_single_succ_edge.
26699 * ipa-utils.c (ipa_merge_profiles): Update.
26700 * loop-doloop.c (add_test): Update.
26701 (doloop_modify): Update.
26702 * loop-unroll.c (compare_and_jump_seq): Update.
26703 (unroll_loop_runtime_iterations): Update.
26704 * lra-constraints.c (lra_inheritance): Update.
26705 * lto-streamer-in.c (input_cfg): Update.
26706 * lto-streamer-out.c (output_cfg): Update.
26707 * mcf.c (adjust_cfg_counts): Update.
26708 * modulo-sched.c (sms_schedule): Update.
26709 * omp-expand.c (expand_omp_for_init_counts): Update.
26710 (extract_omp_for_update_vars): Update.
26711 (expand_omp_ordered_sink): Update.
26712 (expand_omp_for_ordered_loops): Update.
26713 (expand_omp_for_generic): Update.
26714 (expand_omp_for_static_nochunk): Update.
26715 (expand_omp_for_static_chunk): Update.
26716 (expand_cilk_for): Update.
26717 (expand_omp_simd): Update.
26718 (expand_omp_taskloop_for_outer): Update.
26719 (expand_omp_taskloop_for_inner): Update.
26720 * omp-simd-clone.c (simd_clone_adjust): Update.
26721 * optabs.c (expand_doubleword_shift): Update.
26722 (expand_abs): Update.
26723 (emit_cmp_and_jump_insn_1): Update.
26724 (expand_compare_and_swap_loop): Update.
26725 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
26726 * predict.c (predictable_edge_p): Update.
26727 (edge_probability_reliable_p): Update.
26728 (set_even_probabilities): Update.
26729 (combine_predictions_for_insn): Update.
26730 (combine_predictions_for_bb): Update.
26731 (propagate_freq): Update.
26732 (estimate_bb_frequencies): Update.
26733 (force_edge_cold): Update.
26734 * profile-count.c (profile_count::dump): Add missing space into dump.
26735 (profile_count::debug): Add newline.
26736 (profile_count::differs_from_p): Explicitly convert to unsigned.
26737 (profile_count::stream_in): Update.
26738 (profile_probability::dump): New member function.
26739 (profile_probability::debug): New member function.
26740 (profile_probability::differs_from_p): New member function.
26741 (profile_probability::differs_lot_from_p): New member function.
26742 (profile_probability::stream_in): New member function.
26743 (profile_probability::stream_out): New member function.
26744 * profile-count.h (profile_count_quality): Rename to ...
26745 (profile_quality): ... this one.
26746 (profile_probability): New.
26747 (profile_count): Update.
26748 * profile.c (compute_branch_probabilities): Update.
26749 * recog.c (peep2_attempt): Update.
26750 * sched-ebb.c (schedule_ebbs): Update.
26751 * sched-rgn.c (find_single_block_region): Update.
26752 (compute_dom_prob_ps): Update.
26753 (schedule_region): Update.
26754 * sel-sched-ir.c (compute_succs_info): Update.
26755 * stmt.c (struct case_node): Update.
26756 (do_jump_if_equal): Update.
26757 (get_outgoing_edge_probs): Update.
26758 (conditional_probability): Update.
26759 (emit_case_dispatch_table): Update.
26760 (expand_case): Update.
26761 (expand_sjlj_dispatch_table): Update.
26762 (emit_case_nodes): Update.
26763 * targhooks.c: Update.
26764 * tracer.c (better_p): Update.
26765 (find_best_successor): Update.
26766 * trans-mem.c (expand_transaction): Update.
26767 * tree-call-cdce.c: Update.
26768 * tree-cfg.c (gimple_split_edge): Upate.
26769 (move_sese_region_to_fn): Upate.
26770 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
26771 * tree-eh.c (lower_resx): Upate.
26772 (cleanup_empty_eh_move_lp): Upate.
26773 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26774 * tree-inline.c (copy_edges_for_bb): Update.
26775 (copy_cfg_body): Update.
26776 * tree-parloops.c (gen_parallel_loop): Update.
26777 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
26778 (gimple_gen_time_profiler): Update.
26779 * tree-ssa-dce.c (remove_dead_stmt): Update.
26780 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26781 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
26782 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
26783 (unloop_loops): Update.
26784 (try_peel_loop): Update.
26785 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26786 * tree-ssa-loop-split.c (connect_loops): Update.
26787 (split_loop): Update.
26788 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26789 (hoist_guard): Update.
26790 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
26791 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
26792 (value_replacement): Update.
26793 * tree-ssa-reassoc.c (branch_fixup): Update.
26794 * tree-ssa-tail-merge.c (replace_block_by): Update.
26795 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
26796 (create_edge_and_update_destination_phis): Update.
26797 (compute_path_counts): Update.
26798 (recompute_probabilities): Update.
26799 (update_joiner_offpath_counts): Update.
26800 (freqs_to_counts_path): Update.
26801 (duplicate_thread_path): Update.
26802 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
26803 (struct switch_conv_info): Update.
26804 (gen_inbound_check): Update.
26805 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26806 (vect_do_peeling): Update.
26807 (vect_loop_versioning): Update.
26808 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26809 (optimize_mask_stores): Update.
26810 * ubsan.c (ubsan_expand_null_ifn): Update.
26811 * value-prof.c (gimple_divmod_fixed_value): Update.
26812 (gimple_divmod_fixed_value_transform): Update.
26813 (gimple_mod_pow2): Update.
26814 (gimple_mod_pow2_value_transform): Update.
26815 (gimple_mod_subtract): Update.
26816 (gimple_mod_subtract_transform): Update.
26817 (gimple_ic): Update.
26818 (gimple_stringop_fixed_value): Update.
26819 (gimple_stringops_transform): Update.
26820 * value-prof.h: Update.
26822 2017-06-29 Carl Love <cel@us.ibm.com>
26824 * config/rs6000/rs6000-c.c: Add support for built-in functions
26825 vector signed int vec_signed (vector float);
26826 vector signed long long vec_signed (vector double);
26827 vector signed int vec_signed2 (vector double, vector double);
26828 vector signed int vec_signede (vector double);
26829 vector signed int vec_signedo (vector double);
26830 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26831 instruction generator.
26832 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26833 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26834 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26836 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26837 vunsignede_v2df): Add define_expands.
26838 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26839 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26840 VEC_UNSIGNEDO): Add definitions.
26841 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26842 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26843 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
26844 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
26845 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
26846 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26847 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26848 vec_unsignede, vec_unsignedo): Add builtin defines.
26849 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26851 * doc/extend.texi: Update the built-in documentation file for the
26852 new built-in functions.
26854 2017-06-29 Richard Biener <rguenther@suse.de>
26856 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26857 reduction chains to LOOP_VINFO_REDUCTIONS.
26858 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26859 SLP reductions after processing reduction chains.
26861 2017-06-29 Nathan Sidwell <nathan@acm.org>
26863 * builtins.c (fold_builtin_FUNCTION): Use
26864 lang_hooks.decl_printable_name.
26866 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26868 PR middle-end/81194
26869 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26870 with only one label.
26871 * stmt.c (expand_case): Assert NCASES is greater than one.
26873 2017-06-29 Richard Biener <rguenther@suse.de>
26875 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26877 (group_case_labels): Likewise.
26878 (find_taken_edge): Push sanity checking on val to workers...
26879 (find_taken_edge_cond_expr): ... here
26880 (find_taken_edge_switch_expr): ... and here, handle cases
26881 with just a default label.
26882 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26883 (group_case_labels): Likewise.
26884 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26885 group_case_labels does anything cleanup the CFG again.
26887 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26889 PR tree-optimization/81196
26890 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26891 exit condition comparing two IVs.
26893 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26895 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26896 profile to the dummy entry at the end of the list of architectures.
26897 * config/arm/arm-cpu-cdata.h: Regenerated.
26899 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26900 Michael Collison <michael.collison@arm.com>
26903 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26905 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26906 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26907 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26908 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26909 of shift when the shift amount is masked with constant equal to
26910 the size of the mode.
26911 * config/aarch64/predicates.md (subreg_lowpart_operator): New
26914 2017-06-29 Martin Liska <mliska@suse.cz>
26916 * config/i386/i386.opt: Change range from [1,5] to [0,5].
26918 2017-06-29 Yury Gribov <tetra2005@gmail.com>
26921 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
26924 (edge_growth_cache_entry::edge_growth_cache_entry): New
26926 (reset_edge_growth_cache): Update to use constructor.
26928 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26930 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
26931 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26932 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26934 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
26936 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
26937 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
26939 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
26941 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
26942 (*-linux-uclibc*): Add t-uclibc tmake_file.
26943 * config/t-musl: New.
26944 * config/t-uclibc: New.
26946 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
26948 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
26950 (gen_comm_data): Emit architectural setting of arch_prof.
26951 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
26953 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
26954 (armv8-m.base, armv8-m.main): Likewise.
26955 * arm-protos.h (arm_build_target): Add profile field.
26956 (arch_option): Likewise.
26957 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
26959 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
26960 arm_active_target.profile.
26962 2017-06-28 Richard Biener <rguenther@suse.de>
26964 PR middle-end/81227
26965 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
26966 TYPE_OVERFLOW_WRAPS.
26967 * match.pd (negate_expr_p): Likewise.
26968 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
26969 fold_build2, not fold_binary.
26971 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26973 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
26974 Convert memory address to Pmode.
26975 (aarch64_print_operand): Assert MEM operands are always Pmode.
26977 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26980 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
26981 Remove redundant if.
26982 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
26983 * config/arm/aarch-common-protos.h
26984 (aarch_forward_to_shift_is_not_shifted_re): Remove.
26985 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
26987 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
26990 * multiple_target.c (create_dispatcher_calls): Set the default
26991 clone to be static, not public.
26993 2017-06-28 Richard Biener <rguenther@suse.de>
26995 * tree-vect-loop.c (vectorizable_reduction): Move special
26996 cond reduction IV var creation ...
26997 (vect_create_epilog_for_reduction): ... here. Remove induction_index
26998 parameter. Use STMT_VINFO_VECTYPE.
26999 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
27002 2017-06-28 Martin Liska <mliska@suse.cz>
27005 * ipa-visibility.c (non_local_p): Handle visibility.
27007 2017-06-28 Martin Liska <mliska@suse.cz>
27010 * common.opt: Add IntegerRange to various options.
27011 * opt-functions.awk (integer_range_info): New function.
27012 * optc-gen.awk: Add integer_range_info to cl_options struct.
27013 * opts-common.c (decode_cmdline_option): Handle
27014 CL_ERR_INT_RANGE_ARG.
27015 (cmdline_handle_error): Likewise.
27016 * opts.c (print_filtered_help): Show valid interval in
27017 when --help is provided.
27018 * opts.h (struct cl_option): Add range_min and range_max fields.
27019 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
27021 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
27023 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
27024 (x * C EQ/NE y * C): New transformation.
27026 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
27028 * genmultilib (combination_space): Accept '+' in option names.
27030 2017-06-28 Martin Liska <mliska@suse.cz>
27033 * asan.c (instrument_derefs): Bail out inner references
27034 that are hard register variables.
27036 2017-06-28 Jakub Jelinek <jakub@redhat.com>
27039 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
27040 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
27042 2017-06-28 Richard Biener <rguenther@suse.de>
27044 * tree-vectorizer.h (vect_get_vec_defs): Remove.
27045 (vect_get_slp_defs): Adjust.
27046 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
27048 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
27050 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
27051 get_initial_defs_for_reduction instead of vect_get_vec_defs.
27052 (vectorizable_reduction): Adjust.
27053 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
27055 (vect_get_slp_defs): Likewise.
27056 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
27057 (vectorizable_bswap): Adjust.
27058 (vectorizable_call): Likewise.
27059 (vectorizable_conversion): Likewise.
27060 (vectorizable_assignment): Likewise.
27061 (vectorizable_shift): Likewise.
27062 (vectorizable_operation): Likewise.
27063 (vectorizable_store): Likewise.
27064 (vectorizable_condition): Likewise.
27065 (vectorizable_comparison): Likewise.
27067 2017-06-28 Michael Collison <michael.collison@arm.com>
27070 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
27072 (arm_gen_movmemqi): Removed unused variable 'i'.
27073 Convert 'for' loop into 'while' loop.
27074 (arm_expand_prologue): Remove last unnecessary set of insn.
27075 (thumb_pop): Remove unused variable 'pushed_words'.
27076 (thumb_exit): Remove last unnecessary set of regs_to_pop.
27078 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27080 * config/s390/predicates.md: Use s390_rel_address_ok_p.
27081 * config/s390/s390-protos.h: Add prototype of
27082 s390_rel_address_ok_p.
27083 * config/s390/s390.c (s390_got_symbol): New function.
27084 (s390_rel_address_ok_p): New function.
27085 (legitimize_pic_address): Use s390_rel_address_ok_p.
27086 (s390_load_got): Use s390_got_symbol.
27087 (s390_option_override): Issue error if
27088 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
27089 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
27091 * config/s390/s390.opt: New option mpic-data-is-text-relative.
27093 2017-06-27 Andrew Pinski <apinski@cavium.com>
27095 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
27096 (X * copysign (1.0, X)): New pattern.
27097 (X * copysign (1.0, -X)): New pattern.
27098 (copysign (-1.0, CST)): New pattern.
27100 2017-06-27 Joseph Myers <joseph@codesourcery.com>
27102 * genmultilib (combination_space): Remove variable.
27103 Validate reuse rules against regular expression for any sequence
27104 of multilib options in any order.
27106 2017-06-27 Michael Collison <michael.collison@arm.com>
27108 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
27109 call aarch64_split_simd_combine.
27110 * (aarch64_combine_internal<mode>): Delete pattern.
27111 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
27112 Allow register and subreg operands.
27114 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27116 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
27117 specific need, just fallback on defaults.
27118 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
27120 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27121 Olivier Hainque <hainque@adacore.com>
27123 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
27125 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
27126 targets. Pick a default if no particular attempt applied.
27127 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
27130 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27132 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
27133 (x86_64-wrs-vxworks7): Likewise.
27135 2017-06-27 Marek Polacek <polacek@redhat.com>
27138 * ubsan.c (instrument_null): Check get_base_address's result for null.
27140 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27142 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
27144 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27146 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
27147 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
27148 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
27149 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
27150 New function types.
27151 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
27152 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
27153 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
27154 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
27155 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
27156 BUILT_IN_FEUPDATEENV): New builtins.
27157 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
27158 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
27159 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
27160 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
27162 (builtin_structptr_types): Adjust size.
27163 * tree.c (builtin_structptr_types): Add four entries.
27165 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27166 Olivier Hainque <hainque@adacore.com>
27168 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
27169 (TLS_SYM): New local macro, forcing reference to __tls__ on
27170 link command lines for VxWorks 7 RTPs, triggering initialization
27172 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
27173 OS features TLS support, true for RTPs on VxWorks 7.
27174 * config/vxworks.c (vxworks_override_options): Setup emutls
27177 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27179 * predict.c (test_prediction_value_range): Use -1U instead of -1
27180 to avoid narrowing conversion warning.
27181 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
27182 to avoid narrowing conversion warning.
27183 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
27185 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
27187 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27189 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
27190 64bit configurations.
27191 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
27192 (SIZE_TYPE): Likewise.
27193 * config/vxworks.c (vxworks_emutls_var_fields): Use
27194 long_unsigned_type_node instead of unsigned_type_node as the offset
27195 field type, which is "pointer" mode in emutls.c.
27197 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27200 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
27202 PR middle-end/81207
27203 * gimple-fold.c (replace_call_with_call_and_fold): Handle
27204 gimple_vuse copying separately from gimple_vdef copying.
27206 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27208 * value-prof.c (free_hist): Remove call to memset and the enclosing if
27211 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27212 Olivier Hainque <hainque@adacore.com>
27214 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
27215 for all vxworks7 targets.
27216 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
27217 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
27218 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
27219 variations for VX6/VX7 and 32/64bits later on in ...
27220 (VXWORKS_LIB_SPEC): Leverage new macros.
27221 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
27222 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
27224 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27226 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
27227 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
27229 2017-06-26 Carl Love <cel@us.ibm.com>
27231 * config/rs6000/rs6000-c.c: Add support for built-in functions
27232 vector bool char vec_reve (vector bool char);
27233 vector signed char vec_reve (vector signed char);
27234 vector unsigned char vec_reve (vector unsigned char);
27235 vector bool int vec_reve (vector bool int);
27236 vector signed int vec_reve (vector signed int);
27237 vector unsigned int vec_reve (vector unsigned int);
27238 vector bool long long vec_reve (vector bool long long);
27239 vector signed long long vec_reve (vector signed long long);
27240 vector unsigned long long vec_reve (vector unsigned long long);
27241 vector bool short vec_reve (vector bool short);
27242 vector signed short vec_reve (vector signed short);
27243 vector double vec_reve (vector double);
27244 vector float vec_reve (vector float);
27245 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
27246 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
27247 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
27248 (altivec_vreve): New pattern.
27249 * config/rs6000/altivec.h (vec_reve): New define.
27250 * doc/extend.texi (vec_rev): Update the built-in documentation file
27251 for the new built-in functions.
27253 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27255 PR tree-optimization/71815
27256 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
27258 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
27260 (slsr_process_phi): Likewise.
27261 (replace_uncond_cands_and_profitable_phis): Don't replace a
27262 multiply candidate with a stride of 1 (copy or cast).
27263 (phi_incr_cost): Call uses_consumed_by_stmt rather than
27265 (lowest_cost_path): Likewise.
27266 (total_savings): Likewise.
27268 2017-06-26 Richard Biener <rguenther@suse.de>
27271 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
27272 Use def_builtin_pure for all gather builtins.
27274 2017-06-26 Richard Biener <rguenther@suse.de>
27276 PR tree-optimization/81203
27277 * tree-tailcall.c (find_tail_calls): Do not move stmts into
27278 non-dominating BBs.
27280 2017-06-26 Marek Polacek <polacek@redhat.com>
27283 * doc/invoke.texi: Document -Wmultistatement-macros.
27285 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
27287 * doc/sourcebuild.texi (ARM-specific attributes): Document new
27288 arm_neon_ok_no_float_abi effective target.
27290 2017-06-26 Richard Biener <rguenther@suse.de>
27292 PR tree-optimization/80928
27293 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
27294 (copy_bbs): Set BB_DUPLICATED flag early.
27295 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
27297 (execute_on_shrinking_pred): Likewise.
27298 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
27299 BB_DUPLICATED blocks.
27300 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
27301 iterate over all PHIs considering removal of *gsi.
27303 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
27305 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
27308 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27310 * config/rs6000/rs6000-string.c: (expand_block_clear,
27311 do_load_for_compare, select_block_compare_mode,
27312 compute_current_alignment, expand_block_compare,
27313 expand_strncmp_align_check, expand_strn_compare,
27314 expand_block_move, rs6000_output_load_multiple)
27315 Move functions related to string/block move/compare
27316 to a separate file.
27317 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
27318 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
27319 for this function which is now used in two files.
27320 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
27321 * config.gcc: Add rs6000-string.o to extra_objs for
27322 targets powerpc*-*-* and rs6000*-*-*.
27324 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
27327 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
27328 32-bit, since indexed is not valid for DImode.
27329 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
27330 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
27331 (define_peephole2 for Altivec d-form load): Add 32-bit support.
27332 (define_peephole2 for Altivec d-form store): Likewise.
27335 * multiple_target.c (create_dispatcher_calls): Only create the
27336 dispatcher call if the function is the default clone of a
27337 versioned function.
27339 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
27341 PR middle-end/80902
27342 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
27343 a call, force the call to not be a tail call.
27345 2017-06-23 Jeff Law <law@redhat.com>
27347 * doc/contrib.texi: Add entry for Steven Pemberton's work on
27350 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
27352 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
27353 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
27354 handling for early expansion of vector shifts (sl,sr,sra,rl).
27355 (builtin_function_type): Add vector shift right instructions
27356 to the unsigned argument list.
27358 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
27360 rtl-optimizatoin/79286
27361 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
27362 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
27363 trap. PIC register plus a const unspec without offset can never trap.
27365 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
27367 * tree.h (builtin_structptr_type): New type.
27368 (builtin_structptr_types): Declare new array.
27369 * tree.c (builtin_structptr_types): New array.
27370 (free_lang_data, build_common_tree_nodes): Use it.
27372 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
27375 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
27378 2017-06-22 Matt Turner <mattst88@gmail.com>
27380 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
27381 Lake models to skylake case. Assume skylake for unknown
27382 models with clflushopt.
27384 2017-06-22 Jeff Law <law@redhat.com>
27386 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
27387 frame sizes that do not satisfy aarch64_uimm12_shift.
27389 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
27391 * profile-count.h (apply_probability,
27392 apply_scale, probability_in): Fix checks for zero.
27394 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27396 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
27397 * doc/cppdiropts.texi (-I @var{dir}): Document it.
27399 2016-06-22 Richard Biener <rguenther@suse.de>
27401 * tree-vect-loop.c (vect_model_reduction_cost): Handle
27402 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
27403 REDUC_MAX_EXPR support.
27404 (vectorizable_reduction): Likewise.
27405 (vect_create_epilog_for_reduction): Likewise.
27407 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
27409 * match.pd (A / (1 << B) -> A >> B): New.
27410 * generic-match-head.c: Include optabs-tree.h.
27411 * gimple-match-head.c: Likewise.
27412 * optabs-tree.h (target_supports_op_p): New.
27413 * optabs-tree.c (target_supports_op_p): New.
27415 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27417 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
27418 $gcc_cv_ld --help output.
27419 (gcc_cv_ld_demangle): Likewise.
27420 (gcc_cv_ld_eh_frame_hdr): Likewise.
27421 (gcc_cv_ld_pie): Likewise.
27422 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
27423 (gcc_cv_ld_buildid): Likewise.
27424 (gcc_cv_ld_sysroot): Likewise.
27425 (ld_bndplt_support): Likewise.
27426 (ld_pushpopstate_support): Likewise.
27427 * configure: Regenerate.
27428 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
27430 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27433 * config/i386/sse.md (round<mode>2): Renumber match_dup and
27434 operands indexes to avoid gap between operands and match_dups.
27436 2017-06-21 Andrew Pinski <apinski@cavium.com>
27438 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
27439 Increment Arith_shift and Arith_shift_reg by 1.
27440 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
27442 * config/aarch64/aarch64.c (thunderx_tunings): Enable
27443 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
27444 (aarch64_strip_extend): Add new argument and test for it.
27445 (aarch64_cheap_mult_shift_p): New function.
27446 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
27447 add a cost if it is true.
27448 Update calls to aarch64_strip_extend.
27449 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
27451 2017-06-21 Andrew Pinski <apinski@cavium.com>
27453 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
27455 (thunderxt88): Likewise.
27456 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
27457 (thunderx_prefetch_tune): New variable.
27458 (thunderx2t99_prefetch_tune): Update for the correct values.
27459 (thunderxt88_tunings): New variable.
27460 (thunderx_tunings): Use thunderx_prefetch_tune instead of
27461 generic_prefetch_tune.
27462 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
27464 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27466 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
27467 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
27468 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
27469 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
27470 (aarch64_atomic_cas<mode>, GPI): Likewise.
27472 2017-06-21 Martin Liska <mliska@suse.cz>
27474 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
27475 statements on cold and hot labels.
27476 * predict.c (tree_estimate_probability_bb): Remove the
27477 prediction from this place.
27479 2017-06-21 Martin Liska <mliska@suse.cz>
27481 PR tree-optimization/79489
27482 * gimplify.c (maybe_add_early_return_predict_stmt): New
27484 (gimplify_return_expr): Call the function.
27485 * predict.c (tree_estimate_probability_bb): Remove handling
27487 * predict.def: Update comment about early return predictor.
27488 * gimple-predict.h (is_gimple_predict): New function.
27489 * predict.def: Change default value of early return to 66.
27490 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
27492 * passes.def: Put pass_strip_predict_hints to the beginning of
27495 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
27497 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
27498 FUNCTION_DECL declarations.
27499 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
27501 (dwaf2out_decl): Likewise.
27502 * godump.c (go_early_global_decl): Skip call to the real debug hook
27503 for FUNCTION_DECL declarations.
27504 * passes.c (rest_of_decl_compilation): Skip call to the
27505 early_global_decl debug hook for FUNCTION_DECL declarations, unless
27506 -fdump-go-spec is passed.
27508 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27510 * config/i386/i386.c (struct builtin_isa): New field pure_p.
27511 Reorder for compactness.
27512 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
27513 (def_builtin_pure, def_builtin_pure2): New functions.
27514 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
27516 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27518 * match.pd (nop_convert): New predicate.
27519 ((A +- CST1) +- CST2): Allow some NOP conversions.
27521 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27524 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
27525 with ctors/dtors if GOVD_SHARED is set.
27527 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27529 * config/aarch64/aarch64.md (movti_aarch64):
27530 Emit mov rather than orr.
27531 (movtf_aarch64): Likewise.
27532 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
27533 Emit mov rather than orr.
27535 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27537 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
27538 Swap alternatives, make integer dup more expensive.
27540 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27542 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
27543 Return true for non-tls symbols.
27545 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
27547 * config/aarch64/aarch64-cores.def (cortex-a55): New.
27548 (cortex-a75): Likewise.
27549 (cortex-a75.cortex-a55): Likewise.
27550 * config/aarch64/aarch64-tune.md: Regenerate.
27551 * doc/invoke.texi (-mtune): Document new values for -mtune.
27553 2017-06-21 Tom de Vries <tom@codesourcery.com>
27555 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
27556 stack_size feature.
27557 (Effective-Target Keywords, Other attributes): Suggest using
27558 dg-add-options stack_size feature to get stack limit in stack_size
27559 effective target documentation.
27561 2017-06-21 Julian Brown <julian@codesourcery.com>
27562 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27564 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
27565 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
27566 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
27568 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
27569 attribute type list for neon_multiply.
27570 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
27571 attribute type list for neon_multiply.
27572 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
27573 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
27574 attribute type list for neon_multiply.
27575 * config/arm/types.md (crypto_pmull): Add.
27576 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
27577 attribute type list.
27579 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
27581 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
27584 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27586 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
27587 to make sure not to dereference a NULL cost_classes_ptr pointer.
27589 2017-06-20 Carl Love <cel@us.ibm.com>
27591 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27592 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
27593 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
27594 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
27595 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
27596 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
27597 VMULOSW): New enum "unspec" values.
27598 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
27599 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
27600 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
27601 altivec_vmulosw): New patterns.
27602 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
27603 VMULOSW): Add definitions.
27605 2017-06-20 Julia Koval <julia.koval@intel.com>
27607 * config/i386/i386.c: Fix rounding expand for new pattern.
27608 * config/i386/subst.md: Fix pattern (parallel -> unspec).
27610 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27612 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
27613 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
27615 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27617 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
27620 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27622 * config/aarch64/aarch64-cores.def: Rearrange to sort by
27623 architecture, then by implementer ID.
27624 * config/aarch64/aarch64-tune.md: Regenerate.
27626 2017-06-20 Richard Biener <rguenther@suse.de>
27628 PR middle-end/81097
27629 * fold-const.c (split_tree): Fold to type before negating.
27631 2017-06-20 David Malcolm <dmalcolm@redhat.com>
27633 * diagnostic-show-locus.c
27634 (selftest::test_fixit_deletion_affecting_newline): New function.
27635 (selftest::diagnostic_show_locus_c_tests): Call it.
27637 2017-06-20 Andreas Schwab <schwab@suse.de>
27640 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
27643 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27645 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
27646 __ARM_FEATURE_COPROC according to support.
27648 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27650 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27651 Rewritten to avoid overflow for > 32-bit pointers.
27654 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
27655 by removing enum keyword.
27656 (ubsan_type_descriptor): Likewise. Formatting fix.
27659 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
27660 splitter): Require TARGET_SSE2 in the condition.
27662 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
27665 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
27666 for doing vector set of SFmode on ISA 3.0.
27667 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
27668 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
27670 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
27671 SFmode value into a V4SF variable that was extracted from another
27672 V4SF variable without converting the element to double precision
27673 and back to single precision vector format.
27674 (vsx_insert_extract_v4sf_p9_2): Likewise.
27676 2017-06-19 Jakub Jelinek <jakub@redhat.com>
27678 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
27679 in UWHI to avoid undefined overflow.
27682 * ubsan.h (enum ubsan_encode_value_phase): New.
27683 (ubsan_encode_value): Change second argument to
27684 enum ubsan_encode_value_phase with default value of
27685 UBSAN_ENCODE_VALUE_GENERIC.
27686 * ubsan.c (ubsan_encode_value): Change second argument to
27687 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
27688 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
27689 create_tmp_var_raw instead of create_tmp_var and use a
27691 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
27692 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
27693 ubsan_encode_value callers.
27696 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
27697 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
27698 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
27700 2017-06-19 Richard Biener <rguenther@suse.de>
27702 PR middle-end/81118
27703 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
27704 estimates if we changed anything.
27706 2017-06-19 Richard Biener <rguenther@suse.de>
27708 PR tree-optimization/80887
27709 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
27710 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
27711 simplified lookups, then reset mprts_hook.
27712 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
27714 (try_to_simplify): Likewise.
27716 2017-06-19 Martin Liska <mliska@suse.cz>
27719 * gimplify.c (gimplify_switch_expr):
27720 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
27722 2017-06-19 Martin Liska <mliska@suse.cz>
27724 * doc/install.texi: Document that PGO runs in 4 stages.
27726 2017-06-19 Martin Liska <mliska@suse.cz>
27729 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
27730 to dispatcher function name.
27731 * multiple_target.c (replace_function_decl): New function.
27732 (create_dispatcher_calls): Redirect both edges and references.
27734 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
27736 * profile-count.c (profile_count::dump): Dump quality.
27737 (profile_count::differs_from_p): Update for unsigned val.
27738 * profile-count.h (profile_count_quality): New enum.
27739 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
27741 2017-06-19 Richard Biener <rguenther@suse.de>
27743 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
27744 struct function as arg.
27745 (estimate_numbers_of_iterations): Export overload with loop arg.
27746 (free_numbers_of_iterations_estimates_loop): Use an overload of
27747 free_numbers_of_iterations_estimates instead.
27748 * tree-cfg.c (remove_bb): Adjust.
27749 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
27750 * tree-parloops.c (gen_parallel_loop): Likewise.
27751 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
27753 (tree_unroll_loops_completely): Likewise.
27754 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
27755 Use an overload instead and export.
27756 (estimated_loop_iterations): Adjust.
27757 (max_loop_iterations): Likewise.
27758 (likely_max_loop_iterations): Likewise.
27759 (estimate_numbers_of_iterations): Take struct function as arg
27761 (loop_exits_before_overflow): Adjust.
27762 (free_numbers_of_iterations_estimates_loop): Use an overload.
27763 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
27764 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
27766 2017-06-19 Richard Biener <rguenther@suse.de>
27769 * ipa-prop.c (find_constructor_constant_at_offset): Handle
27770 RANGE_EXPR conservatively.
27772 2017-06-16 Carl Love <cel@us.ibm.com>
27774 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27775 definitions for vec_float, vec_float2, vec_floato,
27776 vec_floate built-ins.
27777 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
27778 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
27780 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
27781 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
27782 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
27783 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
27784 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
27785 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
27786 vec_floato): Add builtin defines.
27787 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
27788 Update the built-in documentation file for the new built-in
27791 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27793 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
27794 (mthumb): Mark as the negative of -marm.
27796 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27798 * doc/invoke.texi (ARM Options, -mcpu): Document supported
27800 (ARM Options, -mtune): Document that this accepts the same
27801 extension options as -mcpu.
27802 (ARM Options, -mfpu): Document addition of -mfpu=auto.
27804 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27806 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27807 permitted extensions.
27809 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27811 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27812 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27813 (armv8-m.main): Add option +nodsp.
27814 * config/arm/arm-cpu-cdata.h: Regenerated.
27816 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27818 * config/arm/t-fuchsia: New file.
27819 * config.gcc (arm*-*-fuchsia*): Use it.
27821 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27823 * config/arm/t-symbian: Rewrite for new option infrastructure.
27825 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27827 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27828 (MULTILIB_REQUIRED): Likewise.
27830 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27832 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27833 (MULTILIB_RESUE): Likewise.
27834 (MULTILIB_MATCHES): Likewise.
27835 (MULTLIB_REQUIRED): Likewise.
27837 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27839 * config/arm/t-rtems: Rewrite for new option framework.
27841 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27843 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
27844 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
27845 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
27846 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27847 * config/arm/t-multilib: ... here.
27848 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27849 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27850 armv7-a and armv8*-a when A-profile libraries have not been built.
27851 * config/arm/t-rmprofile: Rewrite.
27853 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27855 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27856 with a backslash. Remove the backslash after substituting unescaped
27858 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27860 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27862 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27863 the driver to pass through the default setting of -mfloat-abi.
27864 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27866 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27867 (all_feat_combs): New rule.
27868 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27870 * config/arm/t-aprofile: Rewrite.
27872 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27874 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27875 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27876 fpu is not specified by the user/command-line.
27877 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27878 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27879 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27880 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27881 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27882 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27884 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27886 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27887 * config/arm/t-arm-elf: Rewritten.
27889 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27891 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27892 have some floating-point instructions.
27893 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27894 (TARGET_MAYBE_HARD_FLOAT): New macro.
27895 * config/arm/arm-builtins.c (arm_init_builtins): Use
27896 TARGET_MAYBE_HARD_FLOAT.
27897 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27899 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27901 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27902 (configargs.h): Include it.
27903 (arm_print_hint_for_fpu_option): New function.
27904 (arm_parse_fpu_option): New function.
27905 (candidate_extension): New class.
27906 (arm_canon_for_multilib): New function.
27907 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27908 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27909 (ARCH_CANONICAL_SPECS): New macro.
27910 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
27912 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27914 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
27915 are set after handling multilib fragments. Set target_cpu_default2
27918 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27920 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
27922 (arm*-*-*): Set target_cpu_default2 to a quoted string.
27923 * config/arm/parsecpu.awk (check_cpu): Validate any extension
27925 (check_arch): Likewise.
27926 * config/arm/arm.c (arm_configure_build_target): Handle
27927 TARGET_CPU_DEFAULT being a string constant. Scan any feature
27928 options in the default.
27930 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27932 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
27933 when an option is an alias of another.
27934 * config/arm/parsecpu.awk (optalias): New parser token.
27935 (gen_comm_data): Mark non-alias options as such. Emit entries
27936 for extension aliases.
27937 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
27938 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
27939 (armv6kz, armv6zk, armv6t2): Likewise.
27940 (armv7): Make vfpv3-d16 an alias.
27941 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
27943 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
27944 Sort in canonical order.
27945 (armv8-a): Sort in canonical order.
27946 (armv8.1-a, armv8.2-a): Likewise.
27947 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
27949 (cortex-a9): Sort in canonical order.
27950 * config/arm/arm.c (selftests.h): Include it.
27951 (arm_test_cpu_arch_data): New function.
27952 (arm_run_self_tests): New function.
27953 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
27954 (targetm): Move declaration to the end of the file.
27955 * arm-cpu-cdata.h: Regenerated.
27957 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27959 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
27960 call to target_mode_check describing the type of option passed.
27961 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
27962 (arm_target_thumb_only): Use arm_parse_arch_option_name or
27963 arm_parse_cpu_option_name to match parameters against list of
27965 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
27966 arm_arch_core_flags data structure.
27967 * config/arm/arm-cpu_cdata.h: Regenerated.
27969 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27971 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
27973 (arm_print_hint_for_cpu_option): Likewise.
27974 (arm_print_hint_for_arch_option): Likewise.
27975 (arm_parse_cpu_option_name): Likewise.
27976 (arm_parse_arch_option_name): Likewise.
27977 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
27978 of entries in the all_fpus list.
27979 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
27980 (arm_parse_cpu_option_name): Declare.
27981 (arm_parse_arch_option_name): Declare.
27982 (arm_parse_option_features): Declare.
27983 (arm_intialize_isa): Declare.
27984 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
27986 (gen_comm_data): ... here. Make definitions non-static.
27987 * config/arm/arm-cpu-data.h: Regenerated.
27988 * config/arm/arm-cpu-cdata.h: Regenerated.
27990 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27992 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
27993 (cpu_arch_extension): New structure.
27994 (cpu_arch_option, arch_option, cpu_option): New structures.
27995 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
27996 architecture types.
27997 (gen_data): Generate new format data tables.
27998 * config/arm/arm.c (cpu_tune): New structure.
27999 (cpu_option, processors): Delete.
28000 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
28001 (arm_print_hint_for_cpu_option): ... this and ...
28002 (arm_print_hint_for_arch_option): ... this.
28003 (arm_parse_arch_cpu_name): Delete. Replace with ...
28004 (arm_parse_cpu_option_name): ... this and ...
28005 (arm_parse_arch_option_name): ... this.
28006 (arm_unrecognized_feature): Change type of target parameter to
28008 (arm_parse_arch_cpu_features): Delete. Replace with ...
28009 (arm_parse_option_features): ... this.
28010 (arm_configure_build_target): Rework to use new configuration data
28012 (arm_print_tune_info): Rework for new configuration data tables.
28013 * config/arm/arm-cpu-data.h: Regenerated.
28014 * config/arm/arm-cpu.h: Regenerated.
28016 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28018 * Makefile.in (OBJS): Move sbitmap.o from here ...
28019 (OBJS-libcommon): ... to here.
28021 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28023 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
28024 (ISA_ALL_CRYPTO): New macro.
28025 (ISA_ALL_SIMD): New macro
28026 (ISA_ALL_FP): New macro.
28027 * config/arm/arm.c (fpu_bitlist): Update initializer.
28028 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
28030 (arm9e): Add fpu. Add option for nofp
28031 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
28032 (arm926ej-s, arm1026ej-s): Likewise.
28033 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
28034 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
28035 neon-fp16, neon-vfpv4, nofp and nosimd.
28036 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
28037 (cortex-a8): Add fpu. Add option for nofp.
28038 (cortex-a9): Add fpu. Add options for nosimd and nofp.
28039 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
28040 (cortex-r4f): Add fpu.
28041 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
28042 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
28044 (cortex-r8): Likewise.
28045 (cortex-m4): Add fpu. Add option for nofp.
28046 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
28047 (cortex-a17.cortex-a7): Likewise.
28048 (cortex-a32): Add fpu. Add options for crypto and nofp.
28049 (cortex-a35, cortex-a53): Likewise.
28050 (cortex-a57): Add fpu. Add option for crypto.
28051 (cortex-a72, cortex-a73): Likewise.
28052 (exynos-m1): Likewise.
28053 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
28054 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
28055 (cortex-m33): Add fpu. Add option for nofp.
28056 * config/arm/arm-cpu-cdata.h: Regenerated
28057 * config/arm/arm-cpu-data.h: Regenerated.
28059 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28061 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
28062 (armv5te, armv5tej): Likewise.
28063 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
28064 (armv7): Add options fp and vfpv3-d16.
28065 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
28066 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
28068 (armv7ve): Likewise.
28069 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
28070 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
28071 (armv8-a): Add nocrypto option.
28072 (armv8.1-a, armv8.2-a): Likewise.
28073 (armv8-m.main): add options fp, fp.dp and nofp.
28075 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28077 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
28079 (armv8-a+crc): Delete.
28080 (armv8.1-a): Add options simd, crypto and nofp.
28081 (armv8.2-a): Add options fp16, simd, crypto and nofp.
28082 (armv8.2-a+fp16): Delete.
28083 (armv8-m.main): Add option dsp.
28084 (armv8-m.main+dsp): Delete.
28085 (cortex-a8): Add fpu. Add nofp option.
28086 (cortex-a9): Add fpu. Add nofp and nosimd options.
28087 * config/arm/parsecpu.awk (gen_data): Generate option tables and
28088 link to main cpu and architecture data structures.
28089 (gen_comm_data): Only put isa attributes from the main architecture
28091 (option): New statement for architecture and CPU entries.
28092 * arm.c (struct cpu_option): New structure.
28093 (struct processors): Add entry for options.
28094 (arm_unrecognized_feature): New function.
28095 (arm_parse_arch_cpu_name): Ignore any characters after the first
28097 (arm_parse_arch_cpu_feature): New function.
28098 (arm_configure_build_target): Separate out any CPU and architecture
28099 features and parse separately. Don't error out if -mfpu=auto is
28100 used with only an architecture string.
28101 (arm_print_asm_arch_directives): New function.
28102 (arm_file_start): Call it.
28103 * config/arm/arm-cpu-cdata.h: Regenerated.
28104 * config/arm/arm-cpu-data.h: Likewise.
28105 * config/arm/arm-tables.opt: Likewise.
28107 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28109 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
28110 assembler when it is not -mfpu=auto.
28112 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28114 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
28115 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
28116 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
28117 (ASM_CPU_SPEC): Rewrite.
28118 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
28119 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
28120 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
28121 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
28122 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
28123 copied string is NUL-terminated. Also strip any characters prefixed
28125 (arm_rewrite_selected_arch): New function.
28126 (arm_rewrite_march): New function.
28128 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28130 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
28131 (x_arm_cpu_string, x_arm_tune_string): Likewise.
28132 (march, mcpu, mtune): Convert to string-based options.
28133 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
28134 (arm_parse_arch_cpu_name): New function.
28135 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
28136 identify selected architecture or CPU.
28137 (arm_option_save): New function.
28138 (TARGET_OPTION_SAVE): Redefine.
28139 (arm_option_restore): Restore string options.
28140 (arm_option_print): Print string options.
28142 2017-06-16 Martin Sebor <msebor@redhat.com>
28144 PR tree-optimization/80933
28145 PR tree-optimization/80934
28146 * builtins.c (fold_builtin_3): Do not handle bcmp here.
28147 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
28148 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
28149 (gimple_fold_builtin): Call them.
28151 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28153 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
28154 as unlikely; update profile.
28156 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28158 * predict.c (force_edge_cold): Handle declaring edges impossible
28161 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28163 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
28165 (try_unroll_loop_completely): Fix reporting.
28167 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28169 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
28171 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
28174 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
28175 if given a non-constant argument for an intrinsic which requires a
28178 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28180 * profile.c (compare_freqs): New function.
28181 (branch_prob): Sort edge list.
28182 (find_spanning_tree): Assume that the list is priority sorted.
28184 2017-06-16 Richard Biener <rguenther@suse.de>
28186 PR tree-optimization/81090
28187 * passes.def (pass_record_bounds): Remove.
28188 * tree-pass.h (make_pass_record_bounds): Likewise.
28189 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
28190 make_pass_record_bounds): Likewise.
28191 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
28192 not free niter estimates at the beginning but at the end.
28193 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
28195 2017-06-16 Richard Biener <rguenther@suse.de>
28197 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
28198 initializer to workaround ICE in host GCC 4.8.
28200 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28202 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
28204 (clone_inlined_nodes): Update.
28206 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28208 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
28209 prefetch settings, and enable prefetching by default at -O3.
28211 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28213 * config/aarch64/aarch64.c (aarch64_override_options_internal):
28214 Set flag_prefetch_loop_arrays according to tuning data.
28216 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28218 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
28219 New tune structure.
28220 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
28221 [Unrelated to main purpose of the patch] Place the pointer field last
28222 to enable type checking errors when tune structure are wrongly merged.
28223 * config/aarch64/aarch64.c (generic_prefetch_tune,)
28224 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
28225 (thunderx2t99_prefetch_tune): New tune constants.
28226 (tune_params *_tunings): Update all tunings (no functional change).
28227 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
28228 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
28229 from tunings structures.
28231 2017-06-16 Jakub Jelinek <jakub@redhat.com>
28234 * ubsan.c (instrument_null): Add T argument, use it instead
28235 of computing it based on IS_LHS.
28236 (instrument_object_size): Likewise.
28237 (pass_ubsan::execute): Adjust instrument_null and
28238 instrument_object_size callers to pass gimple_get_lhs or
28239 gimple_assign_rhs1 result to it. Use instrument_null instead of
28240 calling get_base_address and instrument_mem_ref. Handle
28241 aggregate call arguments for object-size sanitization.
28243 2017-06-16 Yury Gribov <tetra2005@gmail.com>
28245 PR tree-optimization/81089
28246 * tree-vrp.c (is_masked_range_test): Validate operands of
28249 2017-06-15 Martin Sebor <msebor@redhat.com>
28252 * dumpfile.c (dump_register): Avoid calling memset to initialize
28253 a class with a default ctor.
28254 * gcc.c (struct compiler): Remove const qualification.
28255 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
28256 * hash-table.h: Ditto.
28257 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
28259 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
28260 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
28262 * params.h (struct param_info): Make struct members non-const.
28263 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
28264 with default initialization.
28265 * vec.h (vec_copy_construct, vec_default_construct): New helper
28267 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
28268 with vec_copy_construct.
28269 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
28270 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
28271 * doc/invoke.texi (-Wclass-memaccess): Document.
28273 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28275 * emit-rtl.h (is_leaf): Update comment about local
28276 register allocator.
28278 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28281 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
28282 for a variable to have a section before checking if the section has a
28284 Set section to.persistent if persistent attribute is set.
28285 Warn if .persistent attribute is used on an automatic variable.
28287 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
28289 PR rtl-optimization/80474
28290 * reorg.c (update_block): Do not ignore instructions in a delay slot.
28292 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
28294 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
28297 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
28299 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
28300 (casesi): Emit bounds checking as RTL.
28301 (casesi_internal_mips16_<mode>): Remove bounds checking.
28303 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
28305 * config/xtensa/xtensa.c (xtensa_option_override): Append
28306 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
28307 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
28308 xtensa_doloop_hooks): Define unconditionally.
28309 (xtensa_reorg_loops): Only call reorg_loops in the presence of
28311 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
28312 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
28313 for it in xtensa_option_override.
28314 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
28315 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
28317 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
28319 * doc/cppopts.texi: Document '-' special value to -MF.
28321 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
28323 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
28324 (cortex_a53_fconst): Likewise.
28325 (cortex_a53_fpmul): Likewise.
28326 (cortex_a53_f_load_64): Likewise.
28327 (cortex_a53_f_load_many): Likewise.
28328 (cortex_a53_advsimd_alu): Likewise.
28329 (cortex_a53_advsimd_alu_q): Likewise.
28330 (cortex_a53_advsimd_mul): Likewise.
28331 (cortex_a53_advsimd_mul_q): Likewise.
28332 (fpmac bypass): Add new bypass for fpmac-fpmac case.
28333 Add missing fmul, r2f_cvt and fconst cases.
28335 2017-06-14 Richard Biener <rguenther@suse.de>
28337 PR middle-end/81088
28338 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
28340 (fold_binary_loc): When associating do not treat pre-existing
28341 TREE_OVERFLOW on literal constants as a reason to allow
28342 TREE_OVERFLOW on associated literal constants.
28344 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28346 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
28347 (MASK_FEATURES): New macro.
28348 * config/sparc/sparc.c (sparc_option_override): Remove the special
28349 handling of -mfpu and generalize it to all MASK_FEATURES switches.
28351 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28353 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
28354 a division of 0 if non-call exceptions are enabled.
28356 2017-06-14 Andrew Pinski <apinski@cavium.com>
28357 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28360 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
28361 Improve vector initialization code gen for only variable case.
28363 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28365 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
28367 2017-06-14 Richard Biener <rguenther@suse.de>
28369 PR tree-optimization/81083
28370 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
28373 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28375 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
28376 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
28377 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
28378 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
28379 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
28380 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
28382 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28384 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
28385 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
28387 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28389 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
28391 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28393 * config/rs6000/t-rtems: Don't handle SPE.
28395 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28397 * config/rs6000/t-linux: Don't handle SPE.
28399 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28401 * config/rs6000/eabispe.h: Delete file.
28403 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28405 * config/rs6000/t-spe: Delete file.
28407 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28409 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
28410 (rs6000_legitimate_offset_address_p): Return false for anything in
28411 V2SImode or V2SFmode.
28413 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28415 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
28416 except V2SF and V2SI. Rearrange the vector modes, and add comments.
28417 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
28419 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
28420 (rs6000_legitimate_offset_address_p): Ditto.
28421 (rs6000_emit_move): Ditto.
28422 (rs6000_init_builtins): Remove V4HI_type_node.
28424 2017-06-13 Martin Liska <mliska@suse.cz>
28427 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
28428 (gate_asan): Likewise.
28429 * asan.h (asan_no_sanitize_address_p): Remove the function.
28430 (sanitize_flags_p): New function.
28431 * builtins.def: Fix coding style.
28432 * common.opt: Use renamed enum value.
28433 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
28434 * doc/extend.texi: Document no_sanitize attribute.
28435 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
28436 to SANITIZE_UNDEFINED_NONDEFAULT.
28437 * gcc.c (sanitize_spec_function): Use the renamed enum value.
28438 * gimple-fold.c (optimize_atomic_compare_exchange_p):
28439 Use sanitize_flags_p.
28440 * gimplify.c (gimplify_function_tree): Likewise.
28441 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
28442 * opts.c (parse_no_sanitize_attribute): New function.
28443 (common_handle_option): Use renamed enum value.
28444 * opts.h (parse_no_sanitize_attribute): Declare.
28445 * tree.c (sanitize_flags_p): New function.
28446 * tree.h: Declared here.
28447 * tsan.c: Use sanitize_flags_p.
28448 * ubsan.c (ubsan_expand_null_ifn): Likewise.
28449 (instrument_mem_ref): Likewise.
28450 (instrument_bool_enum_load): Likewise.
28451 (do_ubsan_in_current_function): Remove the function.
28452 (pass_ubsan::execute): Use sanitize_flags_p.
28453 * ubsan.h: Remove do_ubsan_in_current_function
28454 * tree-cfg.c (print_no_sanitize_attr_value): New function.
28455 (dump_function_to_file): Use it here.
28457 2017-06-13 Martin Jambor <mjambor@suse.cz>
28459 PR tree-optimization/80803
28460 PR tree-optimization/81063
28461 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
28462 (propagate_subaccesses_across_link): Enqueue subtree whenever
28463 necessary instead of relying on the caller.
28465 2017-06-13 Martin Jambor <mjambor@suse.cz>
28467 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
28468 that have a first_link.
28469 (sort_and_splice_var_accesses): Do not check first_link before
28471 (subtree_mark_written_and_enqueue): Likewise.
28472 (propagate_all_subaccesses): Likewise and do not stop at first
28473 parent with a first_link.
28475 2017-06-13 Martin Jambor <mjambor@suse.cz>
28477 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
28480 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28482 * match.pd: New pattern.
28484 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28486 * tree-vrp.c (is_masked_range_test): New function.
28487 (register_edge_assert_for): Determine ranges for
28490 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28492 PR tree-optimization/67328
28493 * fold-const.c (maskable_range_p): New function.
28494 (build_range_check): Generate bittests if possible.
28496 2017-06-13 Martin Liska <mliska@suse.cz>
28498 * gimple-pretty-print.c (dump_probability): Add new argument.
28499 (dump_edge_probability): Dump both probability and count.
28500 (dump_gimple_label): Likewise.
28501 (dump_gimple_bb_header): Likewise.
28503 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
28506 * config/avr/avr-devices.c: Fix indentation.
28507 * config/avr/gen-avr-mmcu-specs.c: Dito.
28509 2017-06-13 Richard Biener <rguenther@suse.de>
28511 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
28512 instead get vector type from stmt_info.
28513 (vectorizable_reduction): Adjust. Remove dead code.
28515 2017-06-13 Richard Biener <rguenther@suse.de>
28517 PR middle-end/81065
28518 * fold-const.c (extract_muldiv_1): Remove bogus distribution
28519 case of C * (x * C2 + C3).
28520 (fold_addr_of_array_ref_difference): Properly fold index difference.
28522 2017-06-12 David S. Miller <davem@davemloft.net>
28525 * config/sparc/sparc.md (return expander): Emit frame blockage if
28526 function uses alloca.
28528 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
28530 * combine.c (make_field_assignment): Check len rather than the mode
28531 precision when calling force_to_mode.
28533 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
28535 Support multilibs and devices that see flash in RAM address range.
28538 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
28539 (avr_mcu_t) <flash_pm_offset>: New field.
28540 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
28541 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
28542 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
28543 (AVR_TINY_PM_OFFSET): Remove macro.
28544 * config/avr/avr.opt (-mshort-calls): New option.
28545 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
28546 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
28547 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
28548 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
28549 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
28550 instead of avr_arch->have_jmp_call.
28551 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
28552 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
28553 avr_arch->flash_pm_offset to define.
28554 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
28555 new field flash_pm_offset. Add entry for avrxmega3.
28556 (avr_texinfo): Add entry for avrxmega3.
28557 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
28558 attiny212, attiny214,
28559 attiny412, attiny414, attiny416, attiny417,
28560 attiny814, attiny816, attiny817,
28561 attiny1614, attiny1616, attiny1617,
28562 attiny3214, attiny3216, attiny3217.
28563 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
28564 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
28565 (avr_print_operand_address) [AVR_TINY]: Same.
28566 (avr_asm_init_sections) <readonly_data_section>: Only patch
28567 callback if avr_arch->flash_pm_offset = 0.
28568 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
28569 for rodata if avr_arch->flash_pm_offset != 0.
28570 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
28571 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
28572 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
28573 (m_options): Append opt_rcall.
28574 (m_dirnames): Append dir_rcall.
28575 * config/avr/t-multilib: Regenerate.
28577 * configure.ac [target=avr]: Check whether avrxmega3 default
28578 linker description file works as needed.
28579 * configure: Regenerate.
28580 * doc/avr-mmcu.texi: Regenerate.
28581 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
28582 <__AVR_ARCH__>: Document avrxmega3 and 103.
28583 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
28584 <__AVR_SHORT_CALLS__>: Document it.
28585 <__AVR_PM_BASE_ADDRESS__>: Document it.
28586 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
28587 (AVR Variable Attributes) <progmem>: Document this is
28588 not needed for avrxmega3.
28589 (AVR Named Address Spaces) <__flash>: Dito.
28591 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
28593 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
28595 2017-06-12 Doug Rupp <rupp@adacore.com>
28597 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
28598 Append vxworks-stdint.h to the tm_file list.
28599 * config/vxworks-stdint.h: New file.
28601 2017-06-12 Martin Liska <mliska@suse.cz>
28603 PR tree-optimization/81041
28604 * tree-profile.c (gimple_gen_ic_func_profiler):
28605 Create an extra BB in profile-generate
28606 (gimple_gen_time_profiler): Likewise.
28608 2017-06-12 Jakub Jelinek <jakub@redhat.com>
28610 PR tree-optimization/81003
28611 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
28612 (update_range_test): Use it instead of force_gimple_operand_gsi.
28614 2017-06-12 Richard Biener <rguenther@suse.de>
28616 PR tree-optimization/81053
28617 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
28618 with backedge value not defined in loop. Simplify def stmt
28621 2017-06-11 Tom de Vries <tom@codesourcery.com>
28624 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
28626 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
28627 nvptx_cannot_force_const_mem.
28629 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28631 * opts.c (finish_options): Move test for flag_split_stack after
28632 it has been initialized.
28634 2017-06-11 Jason Merrill <jason@redhat.com>
28636 * tree.h (id_equal): New.
28637 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
28638 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
28639 instead of strcmp of IDENTIFIER_POINTER.
28641 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28643 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
28644 (mark_all_inlined_calls_cdtor): Fix formating.
28645 (inline_transform): Rescale profile before inlining.
28647 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28649 * cgraph.h (cgraph_edge::clone): Update prototype.
28650 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
28651 (cgraph_node::create_clone): Update.
28652 (cgraph_node::create_version_clone): Update.
28653 * tree-inline.c (copy_bb): Update.
28654 (expand_call_inline): Update.
28656 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28658 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
28659 factored out from ...
28660 (rs6000_emit_prologue): ... here.
28662 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28664 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
28665 factored out from ...
28666 (rs6000_emit_prologue): ... here.
28668 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28670 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
28672 (handle_missing_profiles): Fix computation of tp_first_run.
28673 (counts_to_freqs): Do not touch freqs when count is 0.
28675 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28677 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
28680 2017-06-10 Tom de Vries <tom@codesourcery.com>
28682 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28683 attributes): Document signal effective target.
28685 2017-06-10 Tom de Vries <tom@codesourcery.com>
28687 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28688 Document effective target stack_size.
28690 2017-06-09 David Malcolm <dmalcolm@redhat.com>
28692 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
28693 to the edit_context if they can be auto-applied.
28695 2017-06-9 Ian Lance Taylor <iant@golang.org>
28697 * opts.c (finish_options): If -fsplit-stack, disable implicit
28698 -forder-blocks-and-partition.
28699 * doc/invoke.texi (Optimize Options): Document that when using
28700 -fsplit-stack -forder-blocks-and-partition is not implicitly
28703 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28705 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
28706 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
28707 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
28708 * builtins.def (abort, trap, unreachable): Declare cold.
28709 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
28710 * tree-core.h (ECF_COLD): New.
28711 * tree.c (set_call_expr_flags): Handle ECF_COLD.
28712 (build_common_builtin_nodes): Mark unreachable and abort as cold.
28714 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28716 * predict.c (unlikely_executed_stmt_p): Cleanup.
28718 2017-06-09 Richard Biener <rguenther@suse.de>
28720 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
28721 model if the ref is always written to.
28723 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28725 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
28727 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28729 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
28732 2017-06-09 Tom de Vries <tom@codesourcery.com>
28735 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
28736 "target cannot support label values" when encountering LABEL_REF.
28738 2017-06-09 Martin Liska <mliska@suse.cz>
28740 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
28741 (gimple_gen_ic_func_profiler): Emit direct comparison
28742 of __gcov_indirect_call_callee with NULL.
28743 (gimple_gen_time_profiler): Change probability from
28744 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
28746 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28748 * profile.c (edge_gcov_counts): Turn to pointer.
28749 (compute_branch_probabilities, compute_branch_probabilities): Update.
28750 (branch_prob): Do not clear edge_gcov_count.
28751 * profile.h (edge_gcov_counts): Turn to pointer.
28752 (edge_gcov_count): Update.
28754 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28756 * gimple.h (gimple_check_failed): Mark cold.
28758 2017-06-09 Richard Biener <rguenther@suse.de>
28760 PR tree-optimization/66623
28761 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
28762 refactor check_reduction into two parts, properly computing
28763 whether we have to check reduction validity for outer loop
28766 2017-06-09 Richard Biener <rguenther@suse.de>
28768 PR tree-optimization/79483
28769 * graphite-scop-detection.c (order): New global.
28770 (get_order): Compute bb to order mapping that satisfies code
28771 generation constraints.
28772 (cmp_pbbs): New helper.
28773 (build_scops): Start domwalk at entry block, sort generated
28776 2017-06-09 Richard Biener <rguenther@suse.de>
28778 PR middle-end/81007
28779 * ipa-polymorphic-call.c
28780 (ipa_polymorphic_call_context::restrict_to_inner_class):
28781 Skip FIELD_DECLs with error_mark_node type.
28782 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
28785 2017-06-09 Martin Liska <mliska@suse.cz>
28787 * predict.c (struct branch_predictor): New struct.
28788 (test_prediction_value_range): New test.
28789 (predict_c_tests): New function.
28790 * selftest-run-tests.c (selftest::run_tests): Run the function.
28791 * selftest.h: Declare new tests.
28793 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
28796 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
28797 gen_add3_insn did not fail.
28798 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
28799 r0, construct that number in a temporary reg and add that reg to r0.
28800 If asked to put the result in r0 as well, fail.
28802 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
28804 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28805 for early expansion of vec_eqv.
28807 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28809 PR middle-end/81005
28810 * ubsan.c (instrument_null): Avoid pointless code temporary.
28811 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28813 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28817 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28820 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28823 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28825 * predict.c (unlikely_executed_edge_p): Move ahead.
28826 (probably_never_executed_edge_p): Use it.
28828 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28830 PR middle-end/79988
28831 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28832 gimple_call_builtin_p call.
28834 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28836 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28837 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28838 rtl_check_failed_type2, rtl_check_failed_code1,
28839 rtl_check_failed_code2, rtl_check_failed_code_mode,
28840 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28841 rtvec_check_failed_bounds, rtl_check_failed_flag,
28842 _fatal_insn_not_found, _fatal_insn): Likewise.
28843 * tree.h (tree_contains_struct_check_failed,
28844 tree_check_failed, tree_not_check_failed,
28845 tree_class_check_failed, tree_range_check_failed,
28846 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28847 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28848 tree_operand_check_failed, omp_clause_check_failed,
28849 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28852 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28854 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28855 flag_branch_probabilities.
28856 * ipa-inline.c (edge_badness): Likewise.
28857 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28858 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28859 * predict.c (maybe_hot_frequency_p): Likewise.
28860 (probably_never_executed): Likewise.
28861 * sched-ebb.c (schedule_ebbs): Likewise.
28862 * sched-rgn.c (find_single_block_region): Likewise.
28863 * tracer.c (tail_duplicate): Likewise.
28865 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28867 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28868 longer requires x_flag_profile_use.
28870 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28872 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28873 instead of flag_reorder_blocks_and_partition.
28874 * dbxout.c (dbxout_function_end): Likewise.
28875 * dwarf2out.c (gen_subprogram_die): Likewise.
28876 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28877 * hw-doloop.c (reorg_loops): Likewise.
28878 * varasm.c (assemble_start_function,
28879 assemble_end_function): Likewise.
28880 (decide_function_section): Do not check for
28881 flag_reorder_blocks_and_partition.
28883 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28885 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28887 (chkp_get_hard_register_fake_addr_expr): Ditto.
28888 (chkp_build_addr_expr): Add check for hard reg case.
28889 (chkp_parse_array_and_component_ref): Ditto.
28890 (chkp_find_bounds_1): Ditto.
28891 (chkp_process_stmt): Don't generate bounds store for
28894 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28896 * predict.c (maybe_hot_bb_p): Do not check profile status.
28897 (maybe_hot_edge_p): Likewise.
28898 (probably_never_executed): Check for zero counts even if profile
28900 (unlikely_executed_edge_p): New function.
28901 (unlikely_executed_stmt_p): New function.
28902 (unlikely_executed_bb_p): New function.
28903 (set_even_probabilities): Use unlikely predicates.
28904 (combine_predictions_for_bb): Likewise.
28905 (predict_paths_for_bb): Likewise.
28906 (predict_paths_leading_to_edge): Likewise.
28907 (determine_unlikely_bbs): New function.
28908 (estimate_bb_frequencies): Use it.
28909 (compute_function_frequency): Use zero counts even if profile is
28911 * profile-count.h: Fix typo.
28913 2017-08-08 Julia Koval <julia.koval@intel.com>
28915 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
28916 _mm512_mask_cvtsepi16_storeu_epi8,
28917 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
28918 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
28919 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
28920 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
28921 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
28922 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
28923 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
28924 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
28925 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
28926 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
28927 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
28928 __builtin_ia32_pmovuswb256mem_mask,
28929 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
28930 __builtin_ia32_pmovwb512mem_mask): New builtins.
28932 2017-08-08 Julia Koval <julia.koval@intel.com>
28934 PR target/73350,80862
28935 * config/i386/subst.md (round): Fix round pattern.
28936 * config/i386/i386.c (ix86_erase_embedded_rounding):
28937 Fix erasing rounding for the fixed pattern.
28939 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28941 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
28943 2017-06-08 Martin Liska <mliska@suse.cz>
28945 PR gcov-profile/80911
28946 * gcov.c (block_info::block_info): New constructor.
28948 2017-06-07 Carl Love <cel@us.ibm.com>
28950 * config/rs6000/rs6000-c: The return type of the following
28951 built-in functions was implemented as int not long long. Fix sign
28952 of return value for the unsigned version of vec_mulo and vec_mule.
28953 vector unsigned long long vec_bperm (vector unsigned long long,
28954 vector unsigned char)
28955 vector signed long long vec_mule (vector signed int,
28957 vector unsigned long long vec_mule (vector unsigned int,
28958 vector unsigned int)
28959 vector signed long long vec_mulo (vector signed int,
28961 vector unsigned long long vec_mulo (vector unsigned int,
28962 vector unsigned int)
28963 * doc/extend.texi: Fix the documentation for the built-in
28966 2017-06-07 Carl Love <cel@us.ibm.com>
28969 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
28972 2017-06-07 Carl Love <cel@us.ibm.com>
28974 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
28975 support, Generate doublehv for signed int/float for BE case only.
28977 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
28979 * doc/invoke.texi (mcx16): Rewrite.
28981 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28983 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
28984 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
28985 *mov<mode>_softfloat, and an anonymous splitter): Use
28986 nonimmediate_operand instead of rs6000_nonimmediate_operand.
28988 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28990 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
28992 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
28993 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
28994 (rs6000_debug_reg_global): Adjust.
28995 (rs6000_init_hard_regno_mode_ok): Adjust.
28996 (rs6000_dbx_register_number): Adjust.
28997 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
28998 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
28999 Remove SPE_ACC and SPEFSCR.
29000 (REG_ALLOC_ORDER): Ditto.
29001 (FRAME_POINTER_REGNUM): Change to 111.
29002 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
29003 (REG_CLASS_NAMES): Ditto.
29004 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
29005 (REGISTER_NAMES): Ditto.
29006 (ADDITIONAL_REG_NAMES): Ditto.
29007 (rs6000_reg_names): Ditto.
29008 * config/rs6000/rs6000.md: Renumber some register number
29011 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29013 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
29015 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
29016 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
29018 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
29019 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
29020 Delete the SPE high registers.
29021 (REG_ALLOC_ORDER): Ditto.
29022 (enum reg_class): Remove SPE_HIGH_REGS.
29023 (REG_CLASS_NAMES): Ditto.
29024 (REG_CLASS_CONTENTS): Delete the SPE high registers.
29025 (REGISTER_NAMES): Ditto.
29026 (rs6000_reg_names): Ditto.
29027 * doc/tm.texi.in: Remove SPE as example.
29028 * doc/tm.texi: Regenerate.
29030 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29032 * config/rs6000/8540.md (ppc8540_brinc): Delete.
29033 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
29034 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
29035 * config/rs6000/rs6000.md (type): Remove "brinc".
29037 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29039 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
29040 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
29041 * config/rs6000/linuxspe.h: Delete file.
29042 * config/rs6000/rs6000.md: Don't include spe.md.
29043 * config/rs6000/spe.h: Delete file.
29044 * config/rs6000/spe.md: Delete file.
29045 * config/rs6000/t-rs6000: Remove spe.md.
29047 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29049 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
29050 (reg_or_none500mem_operand): Delete.
29051 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
29052 instead of reg_or_none500mem_operand.
29054 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29056 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29057 handling of SPE flags.
29058 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
29060 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29062 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
29064 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
29065 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
29066 paired_divv2sf3): Similar.
29067 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
29068 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
29069 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
29071 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
29072 Rename the paired_* instruction patterns.
29073 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
29075 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
29076 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
29077 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
29078 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
29079 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
29080 PAIRED_VECTOR_MODE.
29081 (struct machine_function): Delete field spe_insn_chain_scanned_p.
29082 (spe_func_has_64bit_regs_p): Delete.
29083 (spe_expand_predicate_builtin): Delete.
29084 (spe_expand_evsel_builtin): Delete.
29085 (TARGET_DWARF_REGISTER_SPAN): Do not define.
29086 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
29087 (invalid_e500_subreg): Delete.
29088 (rs6000_legitimize_address): Always force_reg op2 as well, for
29089 paired single memory accesses.
29090 (rs6000_member_type_forces_blk): Delete.
29091 (rs6000_spe_function_arg): Delete.
29092 (rs6000_expand_unop_builtin): Delete SPE handling.
29093 (rs6000_expand_binop_builtin): Ditto.
29094 (spe_expand_stv_builtin): Delete.
29095 (bdesc_2arg_spe): Delete.
29096 (spe_expand_builtin): Delete.
29097 (spe_expand_predicate_builtin): Delete.
29098 (spe_expand_evsel_builtin): Delete.
29099 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
29100 (spe_init_builtins): Delete.
29101 (spe_func_has_64bit_regs_p): Delete.
29102 (savres_routine_name): Delete "info" parameter. Adjust callers.
29103 (rs6000_emit_stack_reset): Ditto.
29104 (rs6000_dwarf_register_span): Delete.
29105 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
29106 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
29107 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
29109 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
29111 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
29112 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
29113 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
29114 mulv2sf3, divv2sf3): Delete expanders.
29116 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29118 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
29120 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29122 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
29123 * config/rs6000/rs6000.c: Ditto.
29125 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29127 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
29128 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
29129 comparison_operator.
29131 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29133 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
29134 * config/rs6000/rs6000.opt: Ditto.
29135 * config/rs6000/t-rtems: Ditto.
29137 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29139 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
29140 TARGET_E500_SINGLE by 0, simplify.
29141 * config/rs6000/rs6000.c: Ditto.
29142 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
29143 (spe_build_register_parallel): Delete.
29144 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
29145 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
29146 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
29147 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
29148 (E500_CONVERT): Delete.
29149 * config/rs6000/spe.md: Remove many patterns and all define_constants.
29151 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29153 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
29154 * config/rs6000/dfp.md: Ditto.
29155 (negdd2, *negdd2_fpr): Merge.
29156 (absdd2, *absdd2_fpr): Merge.
29157 (negtd2, *negtd2_fpr): Merge.
29158 (abstd2, *abstd2_fpr): Merge.
29159 * config/rs6000/e500.h: Delete file.
29160 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
29161 TARGET_FPRS by 1 and simplify.
29162 * config/rs6000/rs6000-c.c: Ditto.
29163 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
29164 TARGET_DF_SPE by 0.
29165 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
29167 * config/rs6000/rs6000.md: Ditto.
29168 (floatdidf2, *floatdidf2_fpr): Merge.
29169 (move_from_CR_gt_bit): Delete.
29170 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
29171 (E500_CR_IOR_COMPARE): Delete.
29172 (All patterns that require !TARGET_FPRS): Delete.
29173 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
29175 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29177 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
29179 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29181 * graphds.c (add_edge): Intitialize edge's attached data.
29182 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
29183 pointer parameter. Call pointed function on each edge during
29184 graph traversing. Skip traversing the edge when the function
29186 (graphds_dfs, graphds_scc): Ditto.
29187 (for_each_edge): New parameter. Pass the new parameter to callback
29189 * graphds.h (skip_edge_callback): New function pointer type.
29190 (graphds_dfs, graphds_scc): New function pointer parameter.
29191 (graphds_edge_callback, for_each_edge): New parameter.
29193 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29195 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
29196 out code checking if runtime alias check is possible to below ...
29197 Call the new function.
29198 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
29199 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
29201 2017-06-07 Marek Polacek <polacek@redhat.com>
29204 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
29205 TYPE_OVERFLOW_WRAPS check.
29207 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29209 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
29210 if versioning is required.
29211 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
29212 peeling with the check for versioning.
29214 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29216 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
29217 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
29218 Set true to new parameter if new ssa variable is defined.
29219 (vect_gen_vector_loop_niters): Refactor. Set range information
29220 for the new vector loop bound variable.
29221 (vect_do_peeling): Ditto.
29223 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29225 * tree-affine.c (ssa.h): Include header file.
29226 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
29227 has wrapping overflow behavior.
29229 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29231 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
29233 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29235 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
29236 (tree_to_aff_combination): ... here.
29238 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29240 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
29241 reg_pressure model function.
29242 (ivopts_global_cost_for_size): Delete.
29243 (determine_set_costs, iv_ca_recount_cost): Call new model function
29244 ivopts_estimate_reg_pressure.
29246 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29248 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
29249 expensive than udiv. Remove floating point cases from mod.
29251 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29253 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
29254 Increase idiv cost.
29256 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29258 * config/aarch64/aarch64.md
29259 (copysignsf3): Fix mask generation.
29261 2017-06-07 Jakub Jelinek <jakub@redhat.com>
29263 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
29265 (class dump_manager): Add register_dumps method.
29266 * dumpfile.c: Include langhooks.h.
29267 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
29268 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
29269 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
29270 (dump_manager::dump_register): Start with 512 entries instead of 32.
29271 (dump_manager::register_dumps): New method.
29272 * toplev.c (general_init): Instead of invoking register_dumps
29273 langhook, invoke register_dumps method on the dump manager.
29274 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
29277 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
29279 * doc/md.texi: Clarify the restrictions on a define_insn condition.
29280 Say that # requires an associated define_split to exist, and that
29281 the define_split must be suitable for use after register allocation.
29283 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29285 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
29286 (compute_outgoing_frequencies): Also initialize zero counts.
29287 (find_many_sub_basic_blocks): Do not produce uninitialized profile
29288 around loops; preserve more of profile when nothing changes.
29290 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
29292 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
29294 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
29295 * config/arm/arm-cpu-cdata.h: Regenerate.
29296 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
29297 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
29298 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
29299 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
29300 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
29302 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
29304 * config/arm/t-rmprofile: Likewise.
29305 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
29307 2017-06-06 David S. Miller <davem@davemloft.net>
29310 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
29311 blockage if function uses alloca.
29313 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29315 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
29316 New "uid" fields to hold pretty-print IDs of group and ref.
29317 Memory references are now identified as <group_id>:<ref_id>
29318 instead of using [random] addresses.
29319 (dump_mem_details): Simplify, no functional change.
29320 (dump_mem_ref): Simplify and make output more concise.
29321 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
29322 (find_or_create_group): Initialize group uid.
29323 (record_ref): Initialize ref uid. Improve debug output.
29324 (prune_group_by_reuse, should_issue_prefetch_p,)
29325 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
29326 (mark_nontemporal_store, determine_loop_nest_reuse):
29327 Improve debug output.
29329 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29331 * dbgcnt.def (prefetch): New debug counter.
29332 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
29333 (schedule_prefetches): Stop issueing prefetches if debug counter
29336 2017-06-06 Tom de Vries <tom@codesourcery.com>
29338 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
29339 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
29341 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29343 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
29344 Use aarch64_reg_or_zero predicate for operand 4.
29345 (aarch64_compare_and_swap<mode> define_insn_and_split):
29346 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
29347 (aarch64_store_exclusive<mode>): Likewise for operand 2.
29349 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
29351 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
29352 (arm_compute_save_core_reg_mask): This.
29353 (thumb1_compute_save_reg_mask): Rename into ...
29354 (thumb1_compute_save_core_reg_mask): This.
29355 (arm_compute_save_reg0_reg12_mask): Adapt comment.
29356 (arm_compute_frame_layout): Likewise.
29358 2017-06-06 Richard Biener <rguenther@suse.de>
29360 PR tree-optimization/80974
29361 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
29362 keep or clear leaders SSA info.
29364 2017-06-06 Tom de Vries <tom@codesourcery.com>
29366 * config/nvptx/nvptx.c (split_mode_p): New function.
29367 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
29369 2017-06-06 Tom de Vries <tom@codesourcery.com>
29371 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
29373 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29376 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
29379 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29381 * shrink-wrap.c (handle_simple_exit): Update profile.
29382 (try_shrink_wrapping): Upate profile.
29384 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29386 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
29387 (tree_guess_outgoing_edge_probabilities): New.
29388 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
29389 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
29391 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29393 * ipa-split.c (split_function): Initialize return bb profile.
29395 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29397 * profile.c (compute_branch_probabilities): Also initialize
29398 EXIT_BLOCK profile.
29400 2017-06-06 Richard Biener <rguenther@suse.de>
29402 PR tree-optimization/80928
29403 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
29404 (vect_analyze_loop_operations): Properly guard analysis for
29406 (vect_transform_loop): Likewise.
29407 (vect_analyze_loop_2): Also reset SLP type on PHIs.
29408 (vect_model_induction_cost): Do not cost for pure SLP.
29409 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
29410 of induction in inner loop vectorization.
29411 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
29412 (vect_get_and_check_slp_defs): Handle vect_induction_def.
29413 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
29415 (vect_analyze_slp_cost_1): Cost induction.
29416 (vect_detect_hybrid_slp_stmts): Handle PHIs.
29417 (vect_get_slp_vect_defs): Likewise.
29418 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
29419 (vect_transform_stmt): Handle SLP reductions.
29420 * tree-vectorizer.h (vectorizable_induction): Adjust.
29422 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29424 * config/rs6000/rs6000.c (make_resolver_func): Update
29425 init_lowered_empty_function call.
29427 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
29429 * doc/invoke.texi: Document the -fprofile-abs-path option.
29430 * common.opt (fprofile-abs-path): New option.
29431 * gcov-io.h (gcov_write_filename): Declare.
29432 * gcov-io.c (gcov_write_filename): New function.
29433 * coverage.c (coverage_begin_function): Use gcov_write_filename.
29434 * profile.c (output_location): Likewise.
29436 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29438 * shring-wrap.c: Revert accidental commit.
29440 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
29442 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
29444 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29446 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
29448 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
29449 profile in callgraph edge.
29450 * profile-count.h (apply_probability): If THIS is 0, then result is 0
29451 (apply_scale): Likewise.
29452 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
29453 Also scale profile when inlining function with zero profile.
29454 (initialize_cfun): Update exit block profile even when it is zero.
29455 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
29456 when profile is read.
29458 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29460 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
29461 (CLONE_*): New constants to define the processors we can generate
29462 code for with the target_clone attribute.
29463 (rs6000_clone_map): New array to identify which clone processors
29464 the current program is running on.
29465 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
29466 target_clone attribute.
29467 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
29468 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
29469 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
29470 (cpu_expand_builtin): Add support for target_clone attribute.
29471 (rs6000_valid_attribute_p): Allow "default" attribute.
29472 (get_decl_name): New debug function to simplify printing the
29473 current function name in debugging statements.
29474 (rs6000_clone_priority): New functions to support the target_clone
29475 attribute, and be able to generate code to switch between ISA 2.05
29476 through ISA 3.0 (power6 through power9).
29477 (rs6000_compare_version_priority): Likewise.
29478 (rs6000_get_function_versions_dispatcher): Likewise.
29479 (make_resolver_func): Likewise.
29480 (add_condition_to_bb): Likewise.
29481 (dispatch_function_versions): Likewise.
29482 (rs6000_generate_version_dispatcher_body): Likewise.
29483 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
29484 (fusion_gpr_load_p): Fix a spacing issue.
29485 * doc/extend.texi (Common Function Attributes): Document that the
29486 PowerPC supports the target_clone attribute.
29488 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29490 * config/arm/arm.h: explain F symbol found in description of ARM
29491 register allocation in its legend.
29493 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29495 * config/mips/frame-header-opt.c: Include profile-count.h.
29496 * config/riscv/riscv.c: Include profile-count.h
29498 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29500 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
29502 (sm_set_flag_if_changed): Add bbs field.
29503 (execute_sm_if_changed_flag_set): Pass BBS.
29504 (execute_sm): Update.
29506 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29508 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
29511 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29513 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
29514 (peephole2): New peephole2 to emit the above.
29515 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
29517 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29519 * config/aarch64/aarch64.c (define_peephole2 above
29520 *sub_<shift>_<mode>): New peephole.
29522 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29524 * config/i386/i386.c (make_resolver_func): Update.
29525 * Makefile.in: Add profile-count.h and profile-count.o
29526 * auto-profile.c (afdo_indirect_call): Update to new API.
29527 (afdo_set_bb_count): Update.
29528 (afdo_propagate_edge): Update.
29529 (afdo_propagate_circuit): Update.
29530 (afdo_calculate_branch_prob): Update.
29531 (afdo_annotate_cfg): Update.
29532 * basic-block.h: Include profile-count.h
29533 (struct edge_def): Turn count to profile_count.
29534 (struct basic_block_def): Likewie.
29535 (REG_BR_PROB_BASE): Move to profile-count.h
29536 (RDIV): Move to profile-count.h
29537 * bb-reorder.c (max_entry_count): Turn to profile_count.
29538 (find_traces): Update.
29539 (rotate_loop):Update.
29540 (connect_traces):Update.
29541 (sanitize_hot_paths):Update.
29542 * bt-load.c (migrate_btr_defs): Update.
29543 * cfg.c (RDIV): Remove.
29544 (init_flow): Use alloc_block.
29545 (alloc_block): Uninitialize count.
29546 (unchecked_make_edge): Uninitialize count.
29547 (check_bb_profile): Update.
29548 (dump_edge_info): Update.
29549 (dump_bb_info): Update.
29550 (update_bb_profile_for_threading): Update.
29551 (scale_bbs_frequencies_int): Update.
29552 (scale_bbs_frequencies_gcov_type): Update.
29553 (scale_bbs_frequencies_profile_count): New.
29554 * cfg.h (update_bb_profile_for_threading): Update.
29555 (scale_bbs_frequencies_profile_count): Declare.
29556 * cfgbuild.c (compute_outgoing_frequencies): Update.
29557 (find_many_sub_basic_blocks): Update.
29558 * cfgcleanup.c (try_forward_edges): Update.
29559 (try_crossjump_to_edge): Update.
29560 * cfgexpand.c (expand_gimple_tailcall): Update.
29561 (construct_exit_block): Update.
29562 * cfghooks.c (verify_flow_info): Update.
29563 (dump_bb_for_graph): Update.
29564 (split_edge): Update.
29565 (make_forwarder_block): Update.
29566 (duplicate_block): Update.
29567 (account_profile_record): Update.
29568 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
29569 (get_estimated_loop_iterations): Update.
29570 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
29571 (single_likely_exit): Update.
29572 * cfgloopmanip.c (scale_loop_profile): Update.
29574 (set_zero_probability): Update.
29575 (lv_adjust_loop_entry_edge): Update.
29576 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
29577 (purge_dead_edges): Update.
29578 (rtl_account_profile_record): Update.
29579 * cgraph.c (cgraph_node::create): Uninitialize count.
29580 (symbol_table::create_edge): Uninitialize count.
29581 (cgraph_update_edges_for_call_stmt_node): Update.
29582 (cgraph_edge::dump_edge_flags): Update.
29583 (cgraph_node::dump): Update.
29584 (cgraph_edge::maybe_hot_p): Update.
29585 * cgraph.h: Include profile-count.h
29586 (create_clone), create_edge, create_indirect_edge): Update.
29587 (cgraph_node): Turn count to profile_count.
29588 (cgraph_edge0: Likewise.
29589 (make_speculative, clone): Update.
29590 (create_edge): Update.
29591 (init_lowered_empty_function): Update.
29592 * cgraphclones.c (cgraph_edge::clone): Update.
29593 (duplicate_thunk_for_node): Update.
29594 (cgraph_node::create_clone): Update.
29595 * cgraphunit.c (cgraph_node::analyze): Update.
29596 (cgraph_node::expand_thunk): Update.
29597 * final.c (dump_basic_block_info): Update.
29598 * gimple-streamer-in.c (input_bb): Update.
29599 * gimple-streamer-out.c (output_bb): Update.
29600 * graphite.c (print_global_statistics): Update.
29601 (print_graphite_scop_statistics): Update.
29602 * hsa-brig.c: Include basic-block.h.
29603 * hsa-dump.c: Include basic-block.h.
29604 * hsa-gen.c (T sum_slice): Update.
29605 (convert_switch_statements):Update.
29606 * hsa-regalloc.c: Include basic-block.h.
29607 * ipa-chkp.c (chkp_produce_thunks): Update.
29608 * ipa-cp.c (struct caller_statistics): Update.
29609 (init_caller_stats): Update.
29610 (gather_caller_stats): Update.
29611 (ipcp_cloning_candidate_p): Update.
29612 (good_cloning_opportunity_p): Update.
29613 (get_info_about_necessary_edges): Update.
29614 (dump_profile_updates): Update.
29615 (update_profiling_info): Update.
29616 (update_specialized_profile): Update.
29617 (perhaps_add_new_callers): Update.
29618 (decide_about_value): Update.
29619 (ipa_cp_c_finalize): Update.
29620 * ipa-devirt.c (struct odr_type_warn_count): Update.
29621 (struct decl_warn_count): Update.
29622 (struct final_warning_record): Update.
29623 (possible_polymorphic_call_targets): Update.
29624 (ipa_devirt): Update.
29625 * ipa-fnsummary.c (redirect_to_unreachable): Update.
29626 * ipa-icf.c (sem_function::merge): Update.
29627 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
29628 * ipa-inline.c (compute_uninlined_call_time): Update.
29629 (compute_inlined_call_time): Update.
29630 (want_inline_small_function_p): Update.
29631 (want_inline_self_recursive_call_p): Update.
29632 (edge_badness): Update.
29633 (lookup_recursive_calls): Update.
29634 (recursive_inlining): Update.
29635 (inline_small_functions): Update.
29636 (dump_overall_stats): Update.
29637 (dump_inline_stats): Update.
29638 * ipa-profile.c (ipa_profile_generate_summary): Update.
29639 (ipa_propagate_frequency): Update.
29640 (ipa_profile): Update.
29641 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29642 * ipa-utils.c (ipa_merge_profiles): Update.
29643 * loop-doloop.c (doloop_modify): Update.
29644 * loop-unroll.c (report_unroll): Update.
29645 (unroll_loop_runtime_iterations): Update.
29646 * lto-cgraph.c (lto_output_edge): Update.
29647 (lto_output_node): Update.
29648 (input_node): Update.
29649 (input_edge): Update.
29650 (merge_profile_summaries): Update.
29651 * lto-streamer-in.c (input_cfg): Update.
29652 * lto-streamer-out.c (output_cfg): Update.
29653 * mcf.c (create_fixup_graph): Update.
29654 (adjust_cfg_counts): Update.
29655 (sum_edge_counts): Update.
29656 * modulo-sched.c (sms_schedule): Update.
29657 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
29658 * predict.c (maybe_hot_count_p): Update.
29659 (probably_never_executed): Update.
29660 (dump_prediction): Update.
29661 (combine_predictions_for_bb): Update.
29662 (propagate_freq): Update.
29663 (handle_missing_profiles): Update.
29664 (counts_to_freqs): Update.
29665 (rebuild_frequencies): Update.
29666 (force_edge_cold): Update.
29667 * predict.h: Include profile-count.h
29668 (maybe_hot_count_p, counts_to_freqs): UPdate.
29669 * print-rtl-function.c: Do not include cfg.h
29670 * print-rtl.c: Include basic-block.h
29671 * profile-count.c: New file.
29672 * profile-count.h: New file.
29673 * profile.c (is_edge_inconsistent): Update.
29674 (correct_negative_edge_counts): Update.
29675 (is_inconsistent): Update.
29676 (set_bb_counts): Update.
29677 (read_profile_edge_counts): Update.
29678 (compute_frequency_overlap): Update.
29679 (compute_branch_probabilities): Update; Initialize and deinitialize
29681 (branch_prob): Update.
29682 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
29683 (edge_gcov_count): New.
29684 (bb_gcov_count): New.
29685 * shrink-wrap.c (try_shrink_wrapping): Update.
29686 * tracer.c (better_p): Update.
29687 * trans-mem.c (expand_transaction): Update.
29688 (ipa_tm_insert_irr_call): Update.
29689 (ipa_tm_insert_gettmclone_call): Update.
29690 * tree-call-cdce.c: Update.
29691 * tree-cfg.c (gimple_duplicate_sese_region): Update.
29692 (gimple_duplicate_sese_tail): Update.
29693 (gimple_account_profile_record): Update.
29694 (execute_fixup_cfg): Update.
29695 * tree-inline.c (copy_bb): Update.
29696 (copy_edges_for_bb): Update.
29697 (initialize_cfun): Update.
29698 (freqs_to_counts): Update.
29699 (copy_cfg_body): Update.
29700 (expand_call_inline): Update.
29701 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
29702 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
29703 (try_unroll_loop_completely): Update.
29704 (try_peel_loop): Update.
29705 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
29706 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
29707 * tree-ssa-loop-split.c (connect_loops): Update.
29708 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
29709 * tree-ssa-reassoc.c (branch_fixup): Update.
29710 * tree-ssa-tail-merge.c (replace_block_by): Update.
29711 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
29712 (compute_path_counts): Update.
29713 (update_profile): Update.
29714 (recompute_probabilities): Update.
29715 (update_joiner_offpath_counts): Update.
29716 (estimated_freqs_path): Update.
29717 (freqs_to_counts_path): Update.
29718 (clear_counts_path): Update.
29719 (ssa_fix_duplicate_block_edges): Update.
29720 (duplicate_thread_path): Update.
29721 * tree-switch-conversion.c (case_bit_test_cmp): Update.
29722 (struct switch_conv_info): Update.
29723 * tree-tailcall.c (decrease_profile): Update.
29724 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
29725 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
29726 * value-prof.c (check_counter): Update.
29727 (gimple_divmod_fixed_value): Update.
29728 (gimple_mod_pow2): Update.
29729 (gimple_mod_subtract): Update.
29730 (gimple_ic_transform): Update.
29731 (gimple_stringop_fixed_value): Update.
29732 * value-prof.h (gimple_ic): Update.
29734 2017-06-02 Carl Love <cel@us.ibm.com>
29736 * config/rs6000/rs6000-c: Add support for built-in functions
29737 vector double vec_doublee (vector signed int);
29738 vector double vec_doublee (vector unsigned int);
29739 vector double vec_doublee (vector float);
29740 vector double vec_doubleh (vector signed int);
29741 vector double vec_doubleh (vector unsigned int);
29742 vector double vec_doubleh (vector float);
29743 vector double vec_doublel (vector signed int);
29744 vector double vec_doublel (vector unsigned int);
29745 vector double vec_doublel (vector float);
29746 vector double vec_doubleo (vector signed int);
29747 vector double vec_doubleo (vector unsigned int);
29748 vector double vec_doubleo (vector float);.
29749 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
29750 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
29752 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
29753 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
29754 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
29756 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
29757 vec_doublel, vec_doubleh.
29758 * doc/extend.texi: Update the built-in documentation file for the
29759 new built-in functions.
29761 2017-06-02 David Malcolm <dmalcolm@redhat.com>
29764 * ipa-inline-analysis.c (free_growth_caches): Set
29765 edge_removal_hook_holder to NULL after removing it.
29767 2017-06-02 Sudakshina Das <sudi.das@arm.com>
29769 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
29770 comparision with zero.
29772 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29773 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29774 for early expansion of vec_min and vec_max builtins.
29775 (builtin_function_type): Add min/max unsigned variants to those
29776 identified as having unsigned arguments.
29778 2017-06-02 Olivier Hainque <hainque@adacore.com>
29780 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
29782 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29784 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
29785 Use VALL_F16 iterator rather than VALL.
29787 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29789 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
29790 Emit CBNZ inside loop when doing a strong exchange and comparing
29791 against zero. Generate the CC flags after the loop.
29793 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
29795 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
29796 (dl_section_ref): New.
29797 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
29798 On AIX, append an expression to subtract the size of the
29799 section length to dl_section_ref.
29801 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29803 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29804 for early expansion of vector absolute builtins.
29806 2017-06-02 Richard Biener <rguenther@suse.de>
29808 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29809 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29811 2017-06-02 Richard Biener <rguenther@suse.de>
29813 PR tree-optimization/80948
29814 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29815 stmt order as well.
29817 2017-06-02 Richard Biener <rguenther@suse.de>
29819 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29821 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29822 for inductions as relevant.
29824 2017-06-02 Richard Biener <rguenther@suse.de>
29826 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29827 (vectorizable_induction): ... this. Remove dead code.
29829 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29831 * builtins. (expand_builtin_alloca): Remove second parameter and
29832 infer its value from the first parameter instead.
29833 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29835 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29837 PR rtl-optimization/80903
29838 * loop-doloop.c (add_test): Unshare sequence.
29840 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29842 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
29844 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
29846 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29848 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29849 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29850 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29851 (xlogue_layout::get_stub_rtx): Make static.
29852 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29853 (xlogue_layout::compute_stub_managed_regs): Rename to...
29854 (xlogue_layout::count_stub_managed_regs): ...this.
29855 (xlogue_layout::is_stub_managed_reg): New function.
29856 (xlogue_layout::m_stub_names): Rename to...
29857 (xlogue_layout::s_stub_names): ...this, make static.
29858 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29859 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29860 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29861 xlogue_layout::s_stub_names): Instantiate statics.
29862 (stub_managed_regs): Remove.
29863 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29864 (disable_call_ms2sysv_xlogues): Rename to...
29865 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29866 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29868 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29869 change after reload_completed.
29870 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29872 (ix86_expand_prologue): Likewise.
29873 (ix86_expand_epilogue): Likewise.
29874 (ix86_expand_split_stack_prologue): Likewise.
29875 (ix86_compute_frame_layout): Remove frame parameter ...
29876 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29877 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29879 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29880 (ix86_frame): Move from here ...
29881 * config/i386/i386.h (ix86_frame): ... to here.
29882 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29883 complete ix86_frame data structure instead. Remove some_ld_name.
29885 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29887 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29888 symbols that hold a DECL_VALUE_EXPR.
29890 2017-06-01 Martin Jambor <mjambor@suse.cz>
29892 PR tree-optimization/80898
29893 * tree-sra.c (process_subtree_disqualification): Removed.
29894 (disqualify_candidate): Do not acll
29895 process_subtree_disqualification.
29896 (subtree_mark_written_and_enqueue): New function.
29897 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29898 RHS has been disqualified and re-queue LHS if necessary. Apart
29899 from that, ignore disqualified RHS.
29901 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29903 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29904 address fetch for z10 or later.
29906 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29908 * config/arc/arc.md (tst_movb): Add guard when splitting.
29910 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29912 * config/arc/arc.c (arc_can_eliminate): Test against
29913 arc_frame_pointer_needed.
29915 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29917 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
29918 to prevent store reordering.
29919 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
29920 (type): Add block type.
29921 (stack_tie): Define special instruction to be used in
29924 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29926 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
29927 constraint. It is not valid for the pattern.
29928 (noncommutative_binary_comparison): Likewise.
29930 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29932 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
29935 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29937 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
29938 be used by the reg-alloc.
29940 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29942 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
29943 reg-alloc when having mul64 or mul32x16 instructions.
29944 (mulsidi3): Likewise.
29945 (umulsidi3): Likewise.
29946 (mulsi32x16): New pattern.
29947 (mulsi64): Likewise.
29948 (mulsidi64): Likewise.
29949 (umulsidi64): Likewise.
29950 (MUL32x16_REG): Define.
29951 (mul64_600): Use MUL32x16_REG.
29952 (mac64_600): Likewise.
29953 (umul64_600): Likewise.
29954 (umac64_600): Likewise.
29956 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29958 * config/arc/arc.md (mulsi3_700): Make it commutative.
29960 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
29962 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
29964 (*sign_extendsidi2_insn): Likewise for movstosw.
29966 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29968 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
29969 the type of the input discriminant value. Convert the
29970 discriminant value of signedness vary.
29972 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
29974 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
29975 Add to -Wall section.
29977 2017-06-01 Richard Biener <rguenther@suse.de>
29979 PR middle-end/66313
29980 * fold-const.c (fold_plusminus_mult_expr): If the factored
29981 factor may be zero use a wrapping type for the inner operation.
29982 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
29983 and handle moved defs.
29984 (process_assignment): Properly guard the unary op case. Return a
29985 tri-state indicating that moving the stmt before the call may allow
29986 to continue. Pass through to_move.
29987 (find_tail_calls): Handle moving unrelated defs before
29990 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
29993 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
29994 splitter result in the canonical way.
29996 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
29998 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
29999 also for 32bit target. Update insn attributes.
30000 (zero-extendsidi2 splitter): Allow all registers for operand 1.
30002 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
30004 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
30005 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
30006 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
30007 (_mm_maskz_min_ss): New intrinsics.
30009 2017-05-31 Martin Liska <mliska@suse.cz>
30011 * tree-vect-loop.c (vect_create_epilog_for_reduction):
30012 Change comment style to one we normally use.
30013 (vectorizable_reduction): Likewise.
30014 (vectorizable_induction): Likewise.
30015 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
30016 (vectorizable_call): Likewise.
30017 (vectorizable_simd_clone_call): Likewise.
30018 (vectorizable_conversion): Likewise.
30019 (vectorizable_assignment): Likewise.
30020 (vectorizable_shift): Likewise.
30021 (vectorizable_operation): Likewise.
30022 (vectorizable_store): Likewise.
30023 (vectorizable_load): Likewise.
30024 * tree-vectorizer.h: Likewise.
30026 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
30028 * passes.c (emergency_dump_function): New.
30029 * tree-pass.h (emergency_dump_function): Declare.
30030 * plugin.c (plugins_internal_error_function): Remove.
30031 * plugin.h (plugins_internal_error_function): Remove declaration.
30032 * toplev.c (internal_error_function): New static function. Use it...
30033 (general_init): ...here.
30035 2017-05-31 Graham Markall <graham.markall@embecosm.com>
30037 * config/arc/arc.c (arc_print_operand): Handle constant operands.
30038 (arc_rtx_costs): Add costs for new patterns.
30039 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
30040 * config/arc/predicates.md: Add _1_2_3_operand predicate.
30042 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
30044 * tree-ssa-strlen.c (get_next_strinfo): New function.
30045 (get_stridx_plus_constant): Use it.
30046 (zero_length_string): Likewise.
30047 (adjust_related_strinfos): Likewise.
30048 (adjust_last_stmt): Likewise.
30050 2017-05-31 Richard Biener <rguenther@suse.de>
30053 * config/i386/i386.c (ix86_expand_builtin): Remove assert
30054 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
30056 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
30058 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
30059 loop_vinfo argument and use of dependence distance vectors.
30060 Check instead whether the two references differ only in their
30061 initial value and assume that they have the same alignment if the
30062 difference is a multiple of the vector alignment.
30063 (vect_analyze_data_refs_alignment): Update call accordingly.
30065 2017-05-31 Martin Liska <mliska@suse.cz>
30068 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
30070 2017-05-31 Bin Cheng <bin.cheng@arm.com>
30072 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
30073 (create_intersect_range_checks): Move from ...
30074 * tree-data-ref.c (create_intersect_range_checks_index)
30075 (create_intersect_range_checks): ... to here.
30076 (create_runtime_alias_checks): New function factored from ...
30077 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
30078 here. Call above function.
30079 * tree-data-ref.h (create_runtime_alias_checks): New function.
30081 2017-05-31 Bin Cheng <bin.cheng@arm.com>
30083 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
30084 segment length for dr_b and compute it in wide_int.
30086 2017-05-31 Richard Biener <rguenther@suse.de>
30088 PR tree-optimization/80906
30089 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
30090 and pass through iv_map.
30091 (copy_bb_and_scalar_dependences): Adjust.
30092 (translate_pending_phi_nodes): Likewise.
30093 (copy_loop_close_phi_args): Handle code-generating IVs instead
30096 2017-05-30 David Malcolm <dmalcolm@redhat.com>
30098 * diagnostic-color.c (color_dict): Add "type-diff".
30099 (parse_gcc_colors): Update comment.
30100 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
30101 -fdiagnostics-show-template-tree and -fno-elide-type.
30102 (GCC_COLORS): Add type-diff to example.
30104 (-fdiagnostics-show-template-tree): New.
30105 (-fno-elide-type): New.
30106 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
30107 the pp_format_decoder callback. Call any m_format_postprocessor's
30109 (pretty_printer::pretty_printer): Initialize
30110 m_format_postprocessor.
30111 (pretty_printer::~pretty_printer): Delete any
30112 m_format_postprocessor.
30113 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
30114 (class format_postprocessor): New class.
30115 (struct pretty_printer::format_decoder): Document the new parameters.
30116 (struct pretty_printer::m_format_postprocessor): New field.
30117 * tree-diagnostic.c (default_tree_printer): Update for new
30118 bool and const char ** params.
30119 * tree-diagnostic.h (default_tree_printer): Likewise.
30121 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
30123 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
30124 (lwa_operand): Delete rs6000_gen_cell_microcode test.
30125 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
30126 rs6000_gen_cell_microcode code.
30127 (rs6000_final_prescan_insn): Delete.
30128 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
30129 "warn-cell-microcode" entries.
30130 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
30131 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
30132 throughout. Change cc_reg_not_micro_cr0_operand to
30133 cc_reg_not_cr0_operand throughout.
30134 (*extendhi<mode>2_noload): Delete.
30135 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
30136 (mwarn-cell-microcode): Delete.
30137 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
30138 -mgen-cell-microcode and -mwarn-cell-microcode.
30140 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
30143 * config/i386/constraints.md (Yd): New constraint.
30145 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
30146 and (?Yd, r) alternatives. Update insn attributes.
30147 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
30148 and (?*Yd, r) alternatives. Update insn attributes.
30149 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
30151 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
30153 * gimplify.c (gimplify_modify_expr): Don't create a
30154 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
30157 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
30159 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
30161 2017-05-30 Richard Biener <rguenther@suse.de>
30163 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
30164 and reduc_def fields.
30165 (STMT_VINFO_REDUC_TYPE): New define.
30166 (STMT_VINFO_REDUC_DEF): Likewise.
30167 (vect_force_simple_reduction): Adjust prototype.
30168 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
30169 (vect_is_simple_reduction): Remove check_reduction argument.
30170 (vect_force_simple_reduction): Adjust and set
30171 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30172 (vectorizable_reduction): Do not re-do reduction analysis
30173 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30174 * tree-parloops.c (gather_scalar_reductions): Adjust.
30176 2017-05-30 Richard Biener <rguenther@suse.de>
30178 PR middle-end/80901
30179 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
30182 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
30184 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
30185 Introduce unknown_misalignment parameter and remove vf.
30186 (vect_peeling_hash_get_lowest_cost):
30187 Pass unknown_misalignment parameter.
30188 (vect_enhance_data_refs_alignment):
30189 Fix unsupportable data ref treatment.
30191 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30193 * tree-vect-data-refs.c (vect_get_data_access_cost):
30194 Workaround for SLP handling.
30195 (vect_enhance_data_refs_alignment):
30196 Compute costs for doing no peeling at all, compare to the best
30197 peeling costs so far and avoid peeling if cheaper.
30199 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30201 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
30202 Return peeling info and set costs to zero for unlimited cost
30204 (vect_enhance_data_refs_alignment): Also inspect all datarefs
30205 with unknown misalignment. Compute and costs for unknown
30206 misalignment, compare them to the costs for known misalignment
30207 and choose the cheapest for peeling.
30209 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30211 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
30212 (vect_get_peeling_costs_all_drs): Create function.
30213 (vect_peeling_hash_get_lowest_cost):
30214 Use vect_get_peeling_costs_all_drs.
30215 (vect_peeling_supportable): Create function.
30216 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
30218 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30220 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
30221 DR_HAS_NEGATIVE_STEP.
30222 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
30223 (vect_enhance_data_refs_alignment): Use.
30224 (vect_duplicate_ssa_name_ptr_info): Use.
30225 * tree-vectorizer.h (dr_misalignment): Use.
30226 (known_alignment_for_access_p): Use.
30228 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
30231 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
30233 (has_section_name): New function.
30235 2017-05-30 Martin Liska <mliska@suse.cz>
30238 * auto-profile.c (get_function_decl_from_block): Fix
30241 2017-05-30 Richard Biener <rguenther@suse.de>
30243 PR middle-end/80876
30244 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
30246 2017-05-30 Martin Liska <mliska@suse.cz>
30248 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
30249 * dumpfile.h (struct dump_file_info): Remove ctors.
30251 2017-05-30 Martin Liska <mliska@suse.cz>
30253 * predict.def: Fix GNU coding style.
30255 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30257 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30258 Mark 'to' argument with ATTRIBUTE_UNUSED.
30260 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30262 * config/xtensa/xtensa.c (xtensa_emit_call): Use
30263 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
30264 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
30267 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
30269 * doc/install.texi (Options specification): Restore entry of
30270 --enable-sjlj-exceptions.
30272 2017-05-27 Michael Eager <eager@eagercon.com>
30275 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
30277 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
30279 * config/microblaze/microblaze.h
30280 (FIXED_REGISTERS): Update in macro.
30281 (CALL_USED_REGISTERS): Update in macro.
30283 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
30285 * doc/install.texi: Add links to macOS binary distributions.
30287 2017-05-27 Jakub Jelinek <jakub@redhat.com>
30291 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30293 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30295 2017-05-26 Martin Liska <mliska@suse.cz>
30297 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
30299 2017-05-26 Martin Liska <mliska@suse.cz>
30301 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
30302 always leading ';; '.
30303 (dump_bb_info): Likewise.
30304 (brief_dump_cfg): Likewise.
30305 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
30306 * dumpfile.c: Remove usage of TDF_VERBOSE.
30307 * dumpfile.h (enum dump_kind): Likewise.
30308 (dump_gimple_bb_header): Do not use TDF_COMMENT.
30309 * print-tree.c (debug_verbose): Remove.
30310 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
30311 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
30312 * tree-diagnostic.c (default_tree_printer): Replace
30313 TDF_DIAGNOSTIC with TDF_SLIM.
30315 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30317 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
30318 in parameter loop, rather than loop_vinfo.
30319 (create_intersect_range_checks): Ditto.
30320 (vect_create_cond_for_alias_checks): Update call to above functions.
30322 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30324 PR tree-optimization/80815
30325 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
30326 for merging runtime alias checks. Handle negative DR_STEPs.
30328 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30330 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
30332 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
30333 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
30334 out code pruning runtime alias checks.
30335 * tree-data-ref.c (prune_runtime_alias_test_list): New function
30336 factored out from above.
30337 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30339 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30341 (prune_runtime_alias_test_list): New decalaration.
30343 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30345 * tree-vect-data-refs.c (compare_tree): Rename and move ...
30346 * tree-data-ref.c (data_ref_compare_tree): ... to here.
30347 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
30348 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
30349 (operator==, comp_dr_with_seg_len_pair): Ditto.
30350 (vect_prune_runtime_alias_test_list): Ditto.
30352 2017-05-26 Martin Liska <mliska@suse.cz>
30355 * params.def: Bound partial-inlining-entry-probability param.
30357 2017-05-26 Marek Polacek <polacek@redhat.com>
30360 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
30363 2017-05-26 Richard Biener <rguenther@suse.de>
30365 PR tree-optimization/80842
30366 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
30369 2017-05-26 Richard Biener <rguenther@suse.de>
30371 PR tree-optimization/80844
30372 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
30374 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
30376 * doc/md.texi (Machine Constraints): Update x86 family
30377 machine constraints section to match 'config/i386/constraints.md'.
30379 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
30381 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
30383 2017-05-25 Nathan Sidwell <nathan@acm.org>
30385 * doc/invoke.texi (--enable-languages): Update documentation.
30387 2017-05-25 Martin Liska <mliska@suse.cz>
30389 * dumpfile.c: Add TDF_FOLDING.
30390 * dumpfile.h (enum dump_kind): Likewise.
30391 * genmatch.c (dt_simplify::gen_1): Use it.
30393 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30395 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
30397 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30399 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30400 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
30402 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30404 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
30405 * match.pd (X == C): Rewrite it here.
30406 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
30407 with_certain_nonzero_bits2): New predicates.
30408 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
30410 2017-05-24 Nathan Sidwell <nathan@acm.org>
30412 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
30415 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
30418 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30420 * config/powerpcspe: New port. Files are copied from the rs6000
30421 port, with "rs6000" in filenames replaced by "powerpcspe".
30423 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
30425 PR rtl-optimization/80754
30426 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
30428 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
30430 * config/sparc/sparc.md (length): Return the correct value for -mflat
30431 sibcalls to match output_sibcall.
30433 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30437 * config/rs6000/rs6000.c (struct machine_function): Add new field
30439 (rs6000_get_separate_components): Init that field, use it.
30440 (rs6000_components_for_bb): Use the field.
30442 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
30444 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
30446 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
30448 PR middle-end/80823
30449 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
30451 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30454 * config/s390/s390.c (s390_check_qrst_address): Check incoming
30455 address against address_operand predicate.
30456 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
30458 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
30460 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
30461 parameters passed indirectly.
30463 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
30465 * config/i386/i386.md (*movdi_internal): Remove SSE4
30466 alternative 18 (?r, *v). Update insn attributes.
30467 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
30468 Update insn attributes.
30469 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
30470 Update insn attributes.
30471 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
30472 alternative 1 (r, v). Remove isa attribute.
30473 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
30474 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
30475 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
30477 2017-05-23 Tom de Vries <tom@codesourcery.com>
30479 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
30482 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30484 * cgraphunit.c (symbol_table::process_new_functions): Update.
30485 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
30486 (inline_generate_summary): Rename to ...
30487 (ipa_fn_summary_generate): ... this one.
30488 (inline_read_summary): Rename to ...
30489 (ipa_fn_summary_read): ... this one.
30490 (inline_write_summary): Rename to ...
30491 (ipa_fn_summary_write): ... this one.
30492 (inline_free_summary): Rename to ...
30493 (ipa_free_fn_summary): ... this one.
30494 (pass_data_local_fn_summary, pass_local_fn_summary,
30495 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
30496 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
30497 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
30498 make_pass_ipa_fn_summary): New.
30499 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
30500 inline_write_summary, inline_free_summary): Remove.
30501 (ipa_free_fn_summary) : New.
30502 * ipa-inline.c (ipa_inline): Update.
30503 (pass_ipa_inline): Do not generate summaries.
30504 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
30506 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
30507 and add pass_ipa_fn_summary.
30508 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
30510 (make_pass_inline_parameters): Remove.
30512 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
30514 * omp-low.c (struct omp_context): Remove "default_kind" member.
30517 * omp-offload.c (execute_oacc_device_lower): Remove the
30518 parallelism dimensions function attributes for unparallelized
30519 OpenACC kernels constructs.
30521 2017-05-23 Martin Liska <mliska@suse.cz>
30523 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
30525 (cgraph_edge::make_speculative): Likewise.
30526 (cgraph_edge::resolve_speculation): Likewise.
30527 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30528 (cgraph_node::dump): Likewise.
30529 * cgraph.h: Likewise.
30530 * cgraphunit.c (analyze_functions): Likewise.
30531 (symbol_table::compile): Likewise.
30532 * ipa-cp.c (print_all_lattices): Likewise.
30533 (determine_versionability): Likewise.
30534 (initialize_node_lattices): Likewise.
30535 (ipcp_verify_propagated_values): Likewise.
30536 (estimate_local_effects): Likewise.
30537 (update_profiling_info): Likewise.
30538 (create_specialized_node): Likewise.
30539 (perhaps_add_new_callers): Likewise.
30540 (decide_about_value): Likewise.
30541 (decide_whether_version_node): Likewise.
30542 (identify_dead_nodes): Likewise.
30543 (ipcp_store_bits_results): Likewise.
30544 * ipa-devirt.c (dump_targets): Likewise.
30545 (ipa_devirt): Likewise.
30546 * ipa-icf.c (sem_item::dump): Likewise.
30547 (sem_function::equals): Likewise.
30548 (sem_variable::equals): Likewise.
30549 (sem_item_optimizer::read_section): Likewise.
30550 (sem_item_optimizer::execute): Likewise.
30551 (congruence_class::dump): Likewise.
30552 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
30553 (dump_inline_summary): Likewise.
30554 (estimate_node_size_and_time): Likewise.
30555 (inline_analyze_function): Likewise.
30556 * ipa-inline-transform.c (inline_call): Likewise.
30557 * ipa-inline.c (report_inline_failed_reason): Likewise.
30558 (want_early_inline_function_p): Likewise.
30559 (edge_badness): Likewise.
30560 (update_edge_key): Likewise.
30561 (inline_small_functions): Likewise.
30562 * ipa-profile.c (ipa_profile): Likewise.
30563 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
30564 (ipa_make_edge_direct_to_target): Likewise.
30565 (remove_described_reference): Likewise.
30566 (ipa_impossible_devirt_target): Likewise.
30567 (propagate_controlled_uses): Likewise.
30568 (ipa_print_node_params): Likewise.
30569 (ipcp_transform_function): Likewise.
30570 * ipa-pure-const.c (pure_const_read_summary): Likewise.
30571 (propagate_pure_const): Likewise.
30572 * ipa-reference.c (generate_summary): Likewise.
30573 (read_write_all_from_decl): Likewise.
30574 (propagate): Likewise.
30575 (ipa_reference_read_optimization_summary): Likewise.
30576 * ipa-utils.c (ipa_merge_profiles): Likewise.
30577 * ipa.c (walk_polymorphic_call_targets): Likewise.
30578 (symbol_table::remove_unreachable_nodes): Likewise.
30579 (ipa_single_use): Likewise.
30580 * passes.c (execute_todo): Likewise.
30581 * predict.c (drop_profile): Likewise.
30582 * symtab.c (symtab_node::get_dump_name): New function.
30583 (symtab_node::dump_name): Likewise.
30584 (symtab_node::dump_asm_name): Likewise.
30585 (symtab_node::dump_references): Likewise.
30586 (symtab_node::dump_referring): Likewise.
30587 (symtab_node::dump_base): Likewise.
30588 (symtab_node::debug_symtab): Likewise.
30589 * tree-sra.c (convert_callers_for_node): Likewise.
30590 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
30591 * value-prof.c (init_node_map): Likewise.
30593 2017-05-23 Martin Liska <mliska@suse.cz>
30595 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
30596 and symtab_node::debug_symtab to symbol_table::debug.
30597 * cgraphunit.c (analyze_functions): Use the renamed function.
30598 (symbol_table::compile): Likewise.
30599 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
30600 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
30601 * passes.c (execute_todo): Likewise.
30602 * symtab.c (symbol_table::dump): New function.
30603 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
30605 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30607 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
30608 that nonconst implies exec.
30610 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30612 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30613 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30614 (inline_edge_summary_vec): Turn into ...
30615 (ipa_call_summaries): ... this one.
30616 (redirect_to_unreachable, edge_set_predicate,
30617 evaluate_properties_for_edge, inline_summary_alloc,
30618 reset_ipa_call_summary, reset_inline_summary,
30619 inline_summary_t::duplicate): Update.
30620 (inline_edge_duplication_hook): Turn to ...
30621 (ipa_call_summary_t::duplicate): ... this one.
30622 (inline_edge_removal_hook): Turn to ...
30623 (ipa_call_summary_t::remove): ... this one.
30624 (dump_inline_edge_summary): Turn to ...
30625 (dump_ipa_call_summary): ... this one.
30626 (estimate_function_body_sizes): Update.
30627 (inline_update_callee_summaries): Update.
30628 (remap_edge_change_prob): Update.
30629 (remap_edge_summaries): Update.
30630 (inline_merge_summary): Update.
30631 (do_estimate_edge_time): Update.
30632 (inline_generate_summary): Update.
30633 (inline_read_section): Update.
30634 (inline_read_summary): Update.
30635 (inline_free_summary): Update.
30636 * ipa-inline.c (can_inline_edge_p): Update.
30637 (compute_inlined_call_time): Update.
30638 (want_inline_small_function_p): Update.
30639 (edge_badness): Update.
30640 (early_inliner): Update.
30641 * ipa-inline.h (inline_edge_summary): Turn to ...
30642 (ipa_call_summary): ... this one.
30643 (ipa_call_summary_t): New class.
30644 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30645 (ipa_call_summaries): New.
30646 (inline_edge_summary): Remove.
30647 (estimate_edge_growth): Update.
30648 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30649 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30650 * ipa-split.c (execute_split_functions): Update.
30651 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30653 2017-05-23 Tom de Vries <tom@codesourcery.com>
30655 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30656 attributes): Document rdrand effective target.
30658 2017-05-23 Tom de Vries <tom@codesourcery.com>
30660 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30661 attributes): Sort alphabetically.
30663 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
30665 * config/avr/genmultilib.awk: Use gsub instead of gensub.
30667 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
30670 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
30671 V2DF/V2DI splat into two separate patterns, one that handles
30672 registers, and the other that only handles memory. Drop support
30673 for splatting from a GPR on ISA 2.07 and then splitting the
30674 splat into direct move and splat.
30675 (vsx_splat_<mode>_reg): Likewise.
30676 (vsx_splat_<mode>_mem): Likewise.
30678 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
30680 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
30682 2017-05-22 Jakub Jelinek <jakub@redhat.com>
30684 PR middle-end/80809
30685 * omp-low.c (finish_taskreg_remap): New function.
30686 (finish_taskreg_scan): If unit size of ctx->record_type
30687 is non-constant, unshare the size expression and replace
30688 decls in it with possible outer var refs.
30690 PR middle-end/80809
30691 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
30692 GOVD_SHARED rather than GOVD_PRIVATE with it.
30693 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
30694 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
30696 PR middle-end/80853
30697 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
30698 as last argument to build_outer_var_ref for pointer bases of array
30699 section reductions.
30701 2017-05-19 Martin Sebor <msebor@redhat.com>
30703 * print-tree.c (print_node): Print DECL_READ_P flag.
30705 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30707 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
30708 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
30709 * cgraph.c: Likewise.
30710 * cgraphunit.c: Likewise.
30711 * gengtype.c: Likewise.
30712 * ipa-cp.c: Likewise.
30713 * ipa-devirt.c: Likewise.
30714 * ipa-icf.c: Likewise.
30715 * ipa-predicate.c: Likewise.
30716 * ipa-profile.c: Likewise.
30717 * ipa-prop.c: Likewise.
30718 * ipa-split.c: Likewise.
30720 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
30721 edge_predicate_pool, dump_inline_hints,
30722 inline_summary::account_size_time, redirect_to_unreachable,
30723 edge_set_predicate, set_hint_predicate,
30724 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
30725 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
30726 inline_summary_t::remove, remap_hint_predicate_after_duplication,
30727 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
30728 ipa_call_summary_t::remove, initialize_growth_caches,
30729 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
30730 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
30731 mark_modified, unmodified_parm_1, unmodified_parm,
30732 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
30733 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
30734 compute_bb_predicates, will_be_nonconstant_expr_predicate,
30735 will_be_nonconstant_predicate, record_modified_bb_info,
30736 get_minimal_bb, record_modified, param_change_prob,
30737 phi_result_unknown_predicate, predicate_for_phi_result,
30738 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
30739 estimate_function_body_sizes, compute_inline_parameters,
30740 compute_inline_parameters_for_curren, pass_data_inline_parameters,
30741 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
30742 inline_update_callee_summaries, remap_edge_change_prob,
30743 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
30744 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
30745 inline_analyze_function, inline_summary_t::insert,
30746 inline_generate_summary, read_ipa_call_summary, inline_read_section,
30747 inline_read_summary, write_ipa_call_summary, inline_write_summary,
30748 inline_free_summary): Move to ipa-fnsummary.h
30749 (predicate_t): Remove.
30750 * ipa-fnsummary.c: New file.
30751 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
30752 (enum inline_hints_vals, inline_hints, agg_position_info,
30753 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
30754 inline_summaries, ipa_call_summary, ipa_call_summary_t,
30755 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
30756 dump_inline_summary, dump_inline_hints, inline_generate_summary,
30757 inline_read_summary, inline_write_summary, inline_free_summary,
30758 inline_analyze_function, initialize_inline_failed,
30759 inline_merge_summary, inline_update_overall_summary,
30760 compute_inline_parameters): Move to ipa-fnsummary.h
30761 * ipa-fnsummary.h: New file.
30762 * ipa-inline-transform.h: Include ipa-inline.h.
30763 * ipa-inline.c: LIkewise.
30765 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30767 * ipa-inline.c (edge_badness): Use inlined_time instead of
30768 inline_summaries->get.
30770 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30772 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
30774 2017-05-22 Nathan Sidwell <nathan@acm.org>
30776 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
30777 (fdump-lang): Document 'raw' option.
30778 * dumpfile.h (TDI_tu): Delete.
30779 * dumpfile.c (dump_files): Remove translation-unit.
30780 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
30782 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
30784 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
30785 command option from $(AWK) call.
30786 * config/avr/genmultilib.awk: Simplify and rewrite so that it
30787 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
30788 [FORMAT]: Remove handling of variable.
30789 * config/avr/t-multilib: Regenerate.
30791 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30793 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
30795 (dump_inline_summary): Do not print self_time.
30796 (estimate_function_body_sizes): Do not set self_time.
30797 (compute_inline_parameters): Likewise.
30798 (inline_read_section, inline_write_summary): Do not stream self_time.
30799 * ipa-inline.h (inline_summary): Drop self_time.
30801 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30803 * ipa-inline-analysis.c (account_size_time): Rename to ...
30804 (inline_summary::account_size_time): ... this one.
30805 (reset_ipa_call_summary): Turn to ...
30806 (ipa_call_summary::reset): ... this one.
30807 (reset_inline_summary): Turn to ...
30808 (inline_summary::reset): ... this one.
30809 (inline_summary_t::remove): Update.
30810 (inline_summary_t::duplicate): Update.
30811 (ipa_call_summary_t::remove): Update.
30812 (dump_inline_summary): Update.
30813 (estimate_function_body_sizes): Update.
30814 (compute_inline_parameters): Update.
30815 (estimate_node_size_and_time): Update.
30816 (inline_merge_summary): Update.
30817 (inline_update_overall_summary): Update.
30818 (inline_read_section): Update.
30819 (inline_write_summary): Update.
30820 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30821 add account_size_time and reset member functions.
30822 (ipa_call_summary): Add reset function.
30823 * ipa-predicate.h (predicate::operator &): Constify.
30825 2017-05-22 Richard Biener <rguenther@suse.de>
30827 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30829 2017-05-19 Jason Merrill <jason@redhat.com>
30831 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30833 2017-05-19 Marek Polacek <polacek@redhat.com>
30836 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30837 TYPE_OVERFLOW_WRAPS checks.
30839 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30841 * tree-core.h (enum omp_clause_default_kind): Add
30842 "OMP_CLAUSE_DEFAULT_PRESENT".
30843 * tree-pretty-print.c (dump_omp_clause): Handle it.
30844 * gimplify.c (enum gimplify_omp_var_data): Add
30845 "GOVD_MAP_FORCE_PRESENT".
30846 (gimplify_adjust_omp_clauses_1): Map it to
30847 "GOMP_MAP_FORCE_PRESENT".
30848 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30850 * gimplify.c (oacc_default_clause): Clarify.
30852 2017-05-19 Nathan Sidwell <nathan@acm.org>
30854 LANG_HOOK_REGISTER_DUMPS
30855 * toplev.c (general_init): Call register dump lang hook.
30856 * doc/invoke.texi: Document -fdump-lang option family.
30857 * dumpfile.c (dump_files): Remove class dump here.
30858 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30859 * dumpfile.h (tree_dump_index): Remove TDI_class.
30860 * langhooks-def.h (lhd_register_dumps): Declare.
30861 (LANG_HOOKS_REGISTER_DUMPS): Define.
30862 (LANG_HOOKS_INITIALIZER): Add it.
30863 * langhooks.c (lhd_register_dumps): Define.
30864 * langhooks.h (struct lang_hooks): Add register_dumps.
30866 2017-05-19 Nathan Sidwell <nathan@acm.org>
30868 * context.h (context::set_passes): New.
30869 * context.c (context::context): Do not create pass manager.
30870 * toplev.c (general_init): Create pass manager here.
30872 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30874 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30875 use this splitter if two add or or instructions would also work for
30876 the constant we want to generate.
30878 2017-05-19 Richard Biener <rguenther@suse.de>
30881 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30882 predicate evaluation.
30884 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30886 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30888 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30891 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30893 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30894 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30895 (inline_edge_summary_vec): Turn into ...
30896 (ipa_call_summaries): ... this one.
30897 (redirect_to_unreachable, edge_set_predicate,
30898 evaluate_properties_for_edge, inline_summary_alloc,
30899 reset_ipa_call_summary, reset_inline_summary,
30900 inline_summary_t::duplicate): Update.
30901 (inline_edge_duplication_hook): Turn to ...
30902 (ipa_call_summary_t::duplicate): ... this one.
30903 (inline_edge_removal_hook): Turn to ...
30904 (ipa_call_summary_t::remove): ... this one.
30905 (dump_inline_edge_summary): Turn to ...
30906 (dump_ipa_call_summary): ... this one.
30907 (estimate_function_body_sizes): Update.
30908 (inline_update_callee_summaries): Update.
30909 (remap_edge_change_prob): Update.
30910 (remap_edge_summaries): Update.
30911 (inline_merge_summary): Update.
30912 (do_estimate_edge_time): Update.
30913 (inline_generate_summary): Update.
30914 (inline_read_section): Update.
30915 (inline_read_summary): Update.
30916 (inline_free_summary): Update.
30917 * ipa-inline.c (can_inline_edge_p): Update.
30918 (compute_inlined_call_time): Update.
30919 (want_inline_small_function_p): Update.
30920 (edge_badness): Update.
30921 (early_inliner): Update.
30922 * ipa-inline.h (inline_edge_summary): Turn to ...
30923 (ipa_call_summary): ... this one.
30924 (ipa_call_summary_t): New class.
30925 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30926 (ipa_call_summaries): New.
30927 (inline_edge_summary): Remove.
30928 (estimate_edge_growth): Update.
30929 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30930 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30931 * ipa-split.c (execute_split_functions): Update.
30932 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30934 2017-05-19 Richard Biener <rguenther@suse.de>
30936 PR middle-end/80764
30937 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
30939 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
30941 * config/rs6000/rs6000.c (struct machine_function): Add field
30942 fpr_is_wrapped_separately.
30943 (rs6000_get_separate_components): Use 64 components. Handle the
30944 new FPR components.
30945 (rs6000_components_for_bb): Handle the FPR components.
30946 (rs6000_emit_prologue_components): Handle the FPR components.
30947 (rs6000_emit_epilogue_components): Handle the FPR components.
30948 (rs6000_set_handled_components): Handle the FPR components.
30949 (rs6000_emit_prologue): Don't output prologue code for those FPRs
30950 that are already separately shrink-wrapped.
30951 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
30952 that are already separately shrink-wrapped.
30954 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30957 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
30960 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
30961 (define_peephole2 for Altivec d-form load): Add peepholes to catch
30962 cases where the register allocator uses a move and an offsettable
30963 memory operation to/from a FPR register on ISA 2.06/2.07.
30964 (define_peephole2 for Altivec d-form store): Likewise.
30966 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
30969 * config/i386/mmx.md (*mov<mode>_internal): Enable
30970 alternatives 11, 12, 13 and 14 also for 32bit targets.
30971 Remove alternatives 15, 16, 17 and 18.
30972 * config/i386/sse.md (vec_concatv2di): Change
30973 alternative (!x, *y) to (x, ?!*Yn).
30975 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
30977 * dumpfile.h (enum dump_kind): Remove stray comma.
30979 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30981 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
30982 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
30983 predicate::num_conditions
30984 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
30985 (CHANGED): turn into predicate::changed.
30986 (agg_position_info): Move to ipa-predicate.h
30987 (add_condition, predicate::add_clause, predicate::operator &=,
30988 predicate::or_with, predicate::evaluate, predicate::probability,
30989 dump_condition, dump_clause, predicate::dump,
30990 predicate::remap_after_duplication, predicate::remap_after_inlining,
30991 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
30992 (evaluate_conditions_for_known_args): Update.
30993 (set_cond_stmt_execution_predicate): Update.
30994 * ipa-inline.h: Include ipa-predicate.h
30995 (condition, inline_param_summary, conditions, agg_position_info,
30996 predicate): Move to ipa-predicate.h
30997 * ipa-predicate.c: New file.
30998 * ipa-predicate.h: New file.
31000 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
31002 * final.c (leaf_function_p): Check we are not in a sequence.
31004 2017-05-18 Martin Liska <mliska@suse.cz>
31006 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
31007 * dumpfile.c (dump_register): Use new enum dump_kind.
31008 (get_dump_file_name): Likewise.
31009 (dump_enable_all): Likewise.
31010 (dump_switch_p_1): Likewise.
31011 (enable_rtl_dump_file): Remove usage of TDF_RTL.
31012 * dumpfile.h (enum dump_kind): New enum type.
31013 (struct dump_file_info): Create constructor and
31014 format fields and comments.
31015 * passes.c (pass_manager::register_one_dump_file):
31017 * statistics.c (statistics_early_init): Likewise.
31018 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
31019 TDF_TREE with TDF_SLIM.
31020 (gather_memory_references_ref): Likewise.
31022 2017-05-18 Martin Liska <mliska@suse.cz>
31024 * vec.h (struct vnull): Use it.
31026 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
31028 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
31029 (true_predicate, false_predicate, true_predicate_p,
31030 false_predicate_p): Remove.
31031 (single_cond_predicate, not_inlined_predicate): Turn to member function
31033 (add_condition): Update.
31034 (add_clause): Turn to...
31035 (predicate::add_clause): ... this one; update; allow passing NULL
31037 (and_predicates): Turn to ...
31038 (predicate::operator &=): ... this one.
31039 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
31040 (or_predicates): Turn to ...
31041 (predicate::or_with): ... this one.
31042 (evaluate_predicate): Turn to ...
31043 (predicate::evaluate): ... this one.
31044 (predicate_probability): Turn to ...
31045 (predicate::probability): ... this one.
31046 (dump_condition): Update.
31047 (dump_predicate): Turn to ...
31048 (predicate::dump): ... this one.
31049 (account_size_time): Update.
31050 (edge_set_predicate): Update.
31051 (set_hint_predicate): UPdate.
31052 (evaluate_conditions_for_known_args): Update.
31053 (evaluate_properties_for_edge): Update.
31054 (remap_predicate_after_duplication): Turn to...
31055 (predicate::remap_after_duplication): ... this one.
31056 (remap_hint_predicate_after_duplication): Update.
31057 (inline_summary_t::duplicate): UPdate.
31058 (dump_inline_edge_summary): Update.
31059 (dump_inline_summary): Update.
31060 (set_cond_stmt_execution_predicate): Update.
31061 (set_switch_stmt_execution_predicate): Update.
31062 (compute_bb_predicates): Update.
31063 (will_be_nonconstant_expr_predicate): Update.
31064 (will_be_nonconstant_predicate): Update.
31065 (phi_result_unknown_predicate): Update.
31066 (predicate_for_phi_result): Update.
31067 (array_index_predicate): Update.
31068 (estimate_function_body_sizes): Update.
31069 (estimate_node_size_and_time): Update.
31070 (estimate_ipcp_clone_size_and_time): Update.
31071 (remap_predicate): Rename to ...
31072 (predicate::remap_after_inlining): ... this one.
31073 (remap_hint_predicate): Update.
31074 (inline_merge_summary): Update.
31075 (inline_update_overall_summary): Update.
31076 (estimate_size_after_inlining): Update.
31077 (read_predicate): Rename to ...
31078 (predicate::stream_in): ... this one.
31079 (read_inline_edge_summary): Update.
31080 (write_predicate): Rename to ...
31081 (predicate::stream_out): ... this one.
31082 (write_inline_edge_summary): Update.
31083 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
31084 (clause_t): Turn to uint32_t
31085 (predicate): Turn to class; implement constructor and operators
31087 (size_time_entry): Update.
31088 (inline_summary): Update.
31089 (inline_edge_summary): Update.
31091 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
31093 * fold-const.c (fold_binary_loc): Move transformation...
31094 * match.pd (C - X CMP X): ... here.
31096 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
31098 * config/sparc/sparc.c (sparc_option_override): Set function
31099 alignment for -mcpu=niagara7 to 64 to match the I$ line.
31100 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
31102 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
31104 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
31106 2017-05-18 Marek Polacek <polacek@redhat.com>
31109 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
31110 (pass_ubsan::execute): Call gimple_assign_single_p instead of
31111 gimple_assign_load_p.
31113 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
31115 PR middle-end/80692
31116 * real.c (do_compare): Give decimal_do_compare preference over
31117 comparing just the signs.
31119 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
31121 * doc/md.texi (Canonicalization of Instructions): Describe the
31122 canonical form of instructions that inherently set a condition
31125 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
31127 PR middle-end/80775
31128 * tree-cfg.c: Move deletion of unreachable case statements to after
31129 the merging of consecutive case labels.
31131 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
31133 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
31134 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
31135 restoring of callee-saved registers.
31137 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
31139 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
31140 * config/visium/visium.c (single_set_and_flags): Likewise.
31141 * config/visium/visium.md (Substitutions): Likewise.
31143 2017-05-17 Martin Liska <mliska@suse.cz>
31145 * cfg.c: Introduce dump_flags_t type and
31146 use it instead of int type.
31148 * cfghooks.c: Likewise.
31149 * cfghooks.h (struct cfg_hooks): Likewise.
31150 * cfgrtl.c: Likewise.
31151 * cfgrtl.h: Likewise.
31152 * cgraph.c (cgraph_node::get_body): Likewise.
31153 * coretypes.h: Likewise.
31154 * domwalk.c: Likewise.
31155 * domwalk.h: Likewise.
31156 * dumpfile.c (struct dump_option_value_info): Likewise.
31157 (dump_enable_all): Likewise.
31158 (dump_switch_p_1): Likewise.
31159 (opt_info_switch_p): Likewise.
31160 * dumpfile.h (enum tree_dump_index): Likewise.
31161 (struct dump_file_info): Likewise.
31162 * genemit.c: Likewise.
31163 * generic-match-head.c: Likewise.
31164 * gengtype.c (open_base_files): Likewise.
31165 * gimple-pretty-print.c: Likewise.
31166 * gimple-pretty-print.h: Likewise.
31167 * graph.c (print_graph_cfg): Likewise.
31168 * graphite-scop-detection.c (dot_all_sese): Likewise.
31169 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
31170 * loop-unroll.c (report_unroll): Likewise.
31171 * passes.c (pass_manager::register_one_dump_file): Likewise.
31172 * print-tree.c: Likewise.
31173 * statistics.c: Likewise.
31174 * tree-cfg.c: Likewise.
31175 * tree-cfg.h: Likewise.
31176 * tree-dfa.c: Likewise.
31177 * tree-dfa.h: Likewise.
31178 * tree-dump.c (dump_function): Likewise.
31179 * tree-dump.h (struct dump_info): Likewise.
31180 * tree-pretty-print.c: Likewise.
31181 * tree-pretty-print.h: Likewise.
31182 * tree-ssa-live.c: Likewise.
31183 * tree-ssa-live.h: Likewise.
31184 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
31185 * tree-vect-loop.c: Likewise.
31186 * tree-vect-slp.c: Likewise.
31188 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
31189 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31191 PR tree-optimization/80457
31192 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
31193 of all arguments to a statement as scalar_to_vec operations.
31194 (vectorizable_call): Adjust call to vect_model_simple_cost for
31196 (vectorizable_conversion): Likewise.
31197 (vectorizable_assignment): Likewise.
31198 (vectorizable_shift): Likewise.
31199 (vectorizable_operation): Likewise.
31200 (vectorizable_comparison): Likewise.
31201 (vect_is_simple_cond): Record the def types for operands.
31202 (vectorizable_condition): Likewise, call vect_model_simple_cost.
31203 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
31204 for statement argument count.
31206 2017-05-16 Carl Love <cel@us.ibm.com>
31208 * config/rs6000/rs6000-c: Add support for built-in functions
31209 vector unsigned long long vec_bperm (vector unsigned long long,
31210 vector unsigned char)
31211 vector signed long long vec_mule (vector signed int,
31213 vector unsigned long long vec_mule (vector unsigned int,
31214 vector unsigned int)
31215 vector signed long long vec_mulo (vector signed int,
31217 vector unsigned long long vec_mulo (vector unsigned int,
31218 vector unsigned int)
31219 vector signed char vec_sldw (vector signed char,
31220 vector signed char,
31222 vector unsigned char vec_sldw (vector unsigned char,
31223 vector unsigned char,
31225 vector signed short vec_sldw (vector signed short,
31226 vector signed short,
31228 vector unsigned short vec_sldw (vector unsigned short,
31229 vector unsigned short,
31231 vector signed int vec_sldw (vector signed int,
31234 vector unsigned int vec_sldw (vector unsigned int,
31235 vector unsigned int,
31237 vector signed long long vec_sldw (vector signed long long,
31238 vector signed long long,
31240 vector unsigned long long vec_sldw (vector unsigned long long,
31241 vector unsigned long long,
31243 * config/rs6000/rs6000-c: Add support for built-in functions
31244 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
31245 * config/rs6000/altivec.h: Add defintion for vec_sldw.
31246 * doc/extend.texi: Update the built-in documentation for the
31247 new built-in functions.
31249 2017-05-16 Marek Polacek <polacek@redhat.com>
31253 * tree.c (save_expr): Don't fold the expression.
31255 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
31257 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
31258 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
31259 and (?*y,m). Update insn attributes.
31261 2017-05-16 Martin Liska <mliska@suse.cz>
31263 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
31264 flags argument of print_gimple_stmt, print_gimple_expr,
31265 print_generic_stmt and print_generic_expr.
31266 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
31267 * coretypes.h: Likewise.
31268 * except.c (dump_eh_tree): Likewise.
31269 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
31270 * gimple-pretty-print.h: Likewise.
31271 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
31272 (backprop::push_to_worklist): Likewise.
31273 (backprop::pop_from_worklist): Likewise.
31274 (backprop::process_use): Likewise.
31275 (backprop::intersect_uses): Likewise.
31276 (note_replacement): Likewise.
31277 * gimple-ssa-store-merging.c
31278 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
31279 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
31280 (pass_store_merging::execute): Likewise.
31281 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
31282 (ssa_base_cand_dump_callback): Likewise.
31283 (dump_incr_vec): Likewise.
31284 (replace_refs): Likewise.
31285 (replace_mult_candidate): Likewise.
31286 (create_add_on_incoming_edge): Likewise.
31287 (create_phi_basis): Likewise.
31288 (insert_initializers): Likewise.
31289 (all_phi_incrs_profitable): Likewise.
31290 (introduce_cast_before_cand): Likewise.
31291 (replace_one_candidate): Likewise.
31292 * gimplify.c (gimplify_expr): Likewise.
31293 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
31294 (set_rename): Likewise.
31295 (rename_uses): Likewise.
31296 (copy_loop_phi_nodes): Likewise.
31297 (add_close_phis_to_merge_points): Likewise.
31298 (copy_loop_close_phi_args): Likewise.
31299 (copy_cond_phi_args): Likewise.
31300 (graphite_copy_stmts_from_block): Likewise.
31301 (translate_pending_phi_nodes): Likewise.
31302 * graphite-poly.c (print_pdr): Likewise.
31303 (dump_gbb_cases): Likewise.
31304 (dump_gbb_conditions): Likewise.
31305 (print_scop_params): Likewise.
31306 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
31307 (build_cross_bb_scalars_use): Likewise.
31308 (gather_bbs::before_dom_children): Likewise.
31309 * hsa-dump.c (dump_hsa_immed): Likewise.
31310 * ipa-cp.c (print_ipcp_constant_value): Likewise.
31311 (get_replacement_map): Likewise.
31312 * ipa-inline-analysis.c (dump_condition): Likewise.
31313 (estimate_function_body_sizes): Likewise.
31314 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
31315 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
31316 * ipa-prop.c (ipa_dump_param): Likewise.
31317 (ipa_print_node_jump_functions_for_edge): Likewise.
31318 (ipa_modify_call_arguments): Likewise.
31319 (ipa_modify_expr): Likewise.
31320 (ipa_dump_param_adjustments): Likewise.
31321 (ipa_dump_agg_replacement_values): Likewise.
31322 (ipcp_modif_dom_walker::before_dom_children): Likewise.
31323 * ipa-pure-const.c (check_stmt): Likewise.
31324 (pass_nothrow::execute): Likewise.
31325 * ipa-split.c (execute_split_functions): Likewise.
31326 * omp-offload.c (dump_oacc_loop_part): Likewise.
31327 (dump_oacc_loop): Likewise.
31328 * trans-mem.c (tm_log_emit): Likewise.
31329 (tm_memopt_accumulate_memops): Likewise.
31330 (dump_tm_memopt_set): Likewise.
31331 (dump_tm_memopt_transform): Likewise.
31332 * tree-cfg.c (gimple_verify_flow_info): Likewise.
31333 (print_loop): Likewise.
31334 * tree-chkp-opt.c (chkp_print_addr): Likewise.
31335 (chkp_gather_checks_info): Likewise.
31336 (chkp_get_check_result): Likewise.
31337 (chkp_remove_check_if_pass): Likewise.
31338 (chkp_use_outer_bounds_if_possible): Likewise.
31339 (chkp_reduce_bounds_lifetime): Likewise.
31340 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
31341 (chkp_mark_completed_bounds): Likewise.
31342 (chkp_register_incomplete_bounds): Likewise.
31343 (chkp_mark_invalid_bounds): Likewise.
31344 (chkp_maybe_copy_and_register_bounds): Likewise.
31345 (chkp_build_returned_bound): Likewise.
31346 (chkp_get_bound_for_parm): Likewise.
31347 (chkp_build_bndldx): Likewise.
31348 (chkp_get_bounds_by_definition): Likewise.
31349 (chkp_generate_extern_var_bounds): Likewise.
31350 (chkp_get_bounds_for_decl_addr): Likewise.
31351 * tree-chrec.c (chrec_apply): Likewise.
31352 * tree-data-ref.c (dump_data_reference): Likewise.
31353 (dump_subscript): Likewise.
31354 (dump_data_dependence_relation): Likewise.
31355 (analyze_overlapping_iterations): Likewise.
31356 * tree-inline.c (expand_call_inline): Likewise.
31357 (tree_function_versioning): Likewise.
31358 * tree-into-ssa.c (dump_defs_stack): Likewise.
31359 (dump_currdefs): Likewise.
31360 (dump_names_replaced_by): Likewise.
31361 (dump_update_ssa): Likewise.
31362 (update_ssa): Likewise.
31363 * tree-object-size.c (pass_object_sizes::execute): Likewise.
31364 * tree-parloops.c (build_new_reduction): Likewise.
31365 (try_create_reduction_list): Likewise.
31366 (ref_conflicts_with_region): Likewise.
31367 (oacc_entry_exit_ok_1): Likewise.
31368 (oacc_entry_exit_single_gang): Likewise.
31369 * tree-pretty-print.h: Likewise.
31370 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
31371 (get_scalar_evolution): Likewise.
31372 (add_to_evolution): Likewise.
31373 (get_loop_exit_condition): Likewise.
31374 (analyze_evolution_in_loop): Likewise.
31375 (analyze_initial_condition): Likewise.
31376 (analyze_scalar_evolution): Likewise.
31377 (instantiate_scev): Likewise.
31378 (number_of_latch_executions): Likewise.
31379 (gather_chrec_stats): Likewise.
31380 (final_value_replacement_loop): Likewise.
31381 (scev_const_prop): Likewise.
31382 * tree-sra.c (dump_access): Likewise.
31383 (disqualify_candidate): Likewise.
31384 (create_access): Likewise.
31385 (reject): Likewise.
31386 (maybe_add_sra_candidate): Likewise.
31387 (create_access_replacement): Likewise.
31388 (analyze_access_subtree): Likewise.
31389 (analyze_all_variable_accesses): Likewise.
31390 (sra_modify_assign): Likewise.
31391 (initialize_constant_pool_replacements): Likewise.
31392 (find_param_candidates): Likewise.
31393 (decide_one_param_reduction): Likewise.
31394 (replace_removed_params_ssa_names): Likewise.
31395 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
31396 * tree-ssa-copy.c (dump_copy_of): Likewise.
31397 (copy_prop_visit_cond_stmt): Likewise.
31398 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
31399 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
31400 (record_equivalences_from_stmt): Likewise.
31401 * tree-ssa-dse.c (compute_trims): Likewise.
31402 (delete_dead_call): Likewise.
31403 (delete_dead_assignment): Likewise.
31404 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
31405 (forward_propagate_into_cond): Likewise.
31406 (pass_forwprop::execute): Likewise.
31407 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
31408 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
31410 (move_computations_worker): Likewise.
31411 (execute_sm): Likewise.
31412 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
31413 (remove_exits_and_undefined_stmts): Likewise.
31414 (remove_redundant_iv_tests): Likewise.
31415 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
31416 (adjust_iv_update_pos): Likewise.
31417 * tree-ssa-math-opts.c (bswap_replace): Likewise.
31418 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
31419 (value_replacement): Likewise.
31420 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
31421 * tree-ssa-pre.c (print_pre_expr): Likewise.
31422 (get_representative_for): Likewise.
31423 (create_expression_by_pieces): Likewise.
31424 (insert_into_preds_of_block): Likewise.
31425 (eliminate_insert): Likewise.
31426 (eliminate_dom_walker::before_dom_children): Likewise.
31427 (eliminate): Likewise.
31428 (remove_dead_inserted_code): Likewise.
31429 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
31430 * tree-ssa-reassoc.c (get_rank): Likewise.
31431 (eliminate_duplicate_pair): Likewise.
31432 (eliminate_plus_minus_pair): Likewise.
31433 (eliminate_not_pairs): Likewise.
31434 (undistribute_ops_list): Likewise.
31435 (eliminate_redundant_comparison): Likewise.
31436 (update_range_test): Likewise.
31437 (optimize_range_tests_var_bound): Likewise.
31438 (optimize_vec_cond_expr): Likewise.
31439 (rewrite_expr_tree): Likewise.
31440 (rewrite_expr_tree_parallel): Likewise.
31441 (linearize_expr): Likewise.
31442 (break_up_subtract): Likewise.
31443 (linearize_expr_tree): Likewise.
31444 (attempt_builtin_powi): Likewise.
31445 (attempt_builtin_copysign): Likewise.
31446 (transform_stmt_to_copy): Likewise.
31447 (transform_stmt_to_multiply): Likewise.
31448 (dump_ops_vector): Likewise.
31449 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
31450 (print_scc): Likewise.
31451 (set_ssa_val_to): Likewise.
31452 (visit_reference_op_store): Likewise.
31453 (visit_use): Likewise.
31454 (sccvn_dom_walker::before_dom_children): Likewise.
31455 (run_scc_vn): Likewise.
31456 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
31458 (expr_hash_elt::print): Likewise.
31459 (const_and_copies::pop_to_marker): Likewise.
31460 (const_and_copies::record_const_or_copy_raw): Likewise.
31461 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
31462 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
31463 (dump_predicates): Likewise.
31464 (find_uninit_use): Likewise.
31465 (warn_uninitialized_phi): Likewise.
31466 (pass_late_warn_uninitialized::execute): Likewise.
31467 * tree-ssa.c (verify_vssa): Likewise.
31468 (verify_ssa): Likewise.
31469 (maybe_optimize_var): Likewise.
31470 * tree-vrp.c (dump_value_range): Likewise.
31471 (dump_all_value_ranges): Likewise.
31472 (dump_asserts_for): Likewise.
31473 (register_edge_assert_for_2): Likewise.
31474 (vrp_visit_cond_stmt): Likewise.
31475 (vrp_visit_switch_stmt): Likewise.
31476 (vrp_visit_stmt): Likewise.
31477 (vrp_visit_phi_node): Likewise.
31478 (simplify_cond_using_ranges_1): Likewise.
31479 (fold_predicate_in): Likewise.
31480 (evrp_dom_walker::before_dom_children): Likewise.
31481 (evrp_dom_walker::push_value_range): Likewise.
31482 (evrp_dom_walker::pop_value_range): Likewise.
31483 (execute_early_vrp): Likewise.
31485 2017-05-16 Richard Biener <rguenther@suse.de>
31487 * dwarf2out.c (loc_list_from_tree_1): Do not create
31488 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
31490 2017-05-16 Richard Biener <rguenther@suse.de>
31492 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
31494 (note_variable_value_in_expr): If we resolved the decl ref
31495 do not push to the stack.
31497 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
31499 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
31500 operations in fast-math mode.
31501 (vaddq_f16): Likewise.
31502 (vmul_f16): Likewise.
31503 (vmulq_f16): Likewise.
31504 (vsub_f16): Likewise.
31505 (vsubq_f16): Likewise.
31506 * config/arm/neon.md (add<mode>3): New.
31508 (fma:<VH:mode>3): New. Also remove outdated comment.
31511 2017-05-16 Martin Liska <mliska@suse.cz>
31515 * ipa-devirt.c (warn_types_mismatch): Fix typo.
31516 (odr_types_equivalent_p): Likewise.
31518 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
31520 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
31522 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
31525 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
31526 non-interunit SSE move alternatives with '?'.
31527 (zero-extendsidi peephole2): New peephole to skip intermediate
31528 general register in SSE zero-extend sequence.
31530 2017-05-15 Jeff Law <law@redhat.com>
31532 * reorg.c (relax_delay_slots): Create a new variable to hold
31533 the temporary target rather than clobbering TARGET_LABEL.
31535 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
31536 missing argument to extract_bit_field call.
31537 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
31539 2017-05-15 Martin Liska <mliska@suse.cz>
31542 * gcc.c (process_command): Do not allow empty argument of -o option.
31544 2017-05-15 Renlin Li <renlin.li@arm.com>
31546 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
31547 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
31548 * config/aarch64/constraints.md (Usf): Add long call check.
31549 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
31550 (call_value): Likewise.
31551 (sibcall): Likewise.
31552 (sibcall_value): Likewise.
31554 (call_value_insn): New.
31555 (sibcall_insn): Update rtx pattern.
31556 (sibcall_value_insn): Likewise.
31557 (call_internal): Remove.
31558 (call_value_internal): Likewise.
31559 (sibcall_internal): Likewise.
31560 (sibcall_value_internal): Likewise.
31561 (call_reg): Likewise.
31562 (call_symbol): Likewise.
31563 (call_value_reg): Likewise.
31564 (call_value_symbol): Likewise.
31566 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
31569 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
31571 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
31573 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
31574 compatible with CCGOCmode and with CCZmode.
31576 2017-05-14 Martin Sebor <msebor@redhat.com>
31578 PR middle-end/77671
31579 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
31580 (gimple_fold_builtin_snprintf): Same.
31581 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
31582 (gimple_fold_builtin_snprintf): Same.
31583 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
31584 of character types.
31585 (is_call_safe): New function.
31586 (try_substitute_return_value): Call it.
31587 (try_simplify_call): New function.
31588 (pass_sprintf_length::handle_gimple_call): Call it.
31590 2017-05-14 Martin Sebor <msebor@redhat.com>
31592 PR middle-end/80669
31593 * builtins.c (expand_builtin_stpncpy): Simplify.
31595 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
31597 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
31598 * config/i386/i386.h
31599 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
31600 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
31601 (struct machine_function): Add new members call_ms2sysv,
31602 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
31603 (struct machine_frame_state): New fields sp_realigned and
31604 sp_realigned_offset.
31605 * config/i386/i386.c
31606 (enum xlogue_stub): New enum.
31607 (enum xlogue_stub_sets): New enum.
31608 (class xlogue_layout): New class.
31609 (struct ix86_frame): New fields stack_realign_allocate_offset,
31610 stack_realign_offset and outlined_save_offset. Modify comments to
31611 detail stack layout when using out-of-line stubs.
31612 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
31613 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
31614 -mcall-ms2sysv-xlogues.
31615 (stub_managed_regs): New static variable.
31616 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
31617 registers managed by out-of-line stub.
31618 (disable_call_ms2sysv_xlogues): New function.
31619 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
31620 m->call_ms2sysv when appropriate and compute frame layout for
31622 (sp_valid_at, fp_valid_at): New inline functions.
31623 (choose_basereg): New function.
31624 (choose_baseaddr): Add align parameter, use choose_basereg and modify
31626 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
31627 Use align parameter of choose_baseaddr to generated aligned SSE movs
31629 (pro_epilogue_adjust_stack): Modify to track
31630 machine_frame_state::sp_realigned.
31631 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
31632 (ix86_nsaved_sseregs): Likewise.
31633 (ix86_emit_save_regs): Likewise.
31634 (ix86_emit_save_regs_using_mov): Likewise.
31635 (ix86_emit_save_sse_regs_using_mov): Likewise.
31636 (get_scratch_register_on_entry): Likewise.
31637 (gen_frame_set): New function.
31638 (gen_frame_load): Likewise.
31639 (gen_frame_store): Likewise.
31640 (emit_outlined_ms2sysv_save): Likewise.
31641 (emit_outlined_ms2sysv_restore): Likewise.
31642 (ix86_expand_prologue): Modify stack re-alignment code and call
31643 emit_outlined_ms2sysv_save when appropriate.
31644 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
31645 parameter rtx_insn *insn, which allows the function to be used to only
31646 generate the notes.
31647 (ix86_expand_epilogue): Modify validity checks of frame and stack
31648 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
31649 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
31650 * config/i386/predicates.md
31651 (save_multiple): New predicate.
31652 (restore_multiple): Likewise.
31653 * config/i386/sse.md
31654 (save_multiple<mode>): New pattern.
31655 (save_multiple_realign<mode>): Likewise.
31656 (restore_multiple<mode>): Likewise.
31657 (restore_multiple_and_return<mode>): Likewise.
31658 (restore_multiple_leave_return<mode>): Likewise.
31659 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
31661 2017-05-14 Julia Koval <julia.koval@intel.com>
31663 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
31664 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
31665 (__builtin_ia32_xsetbv): New builtins.
31666 * config/i386/i386.c (ix86_expand_special_args_builtin):
31668 (ix86_expand_builtin): Special expand for new intrinsics.
31669 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
31670 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
31671 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
31673 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31675 * cfganal.c (inverted_post_order_compute): Change argument type
31677 * cfganal.h (inverted_post_order_compute): Adjust prototype.
31678 * df-core.c (rest_of_handle_df_initialize): Adjust.
31679 (rest_of_handle_df_finish): Likewise.
31680 (df_analyze_1): Likewise.
31681 (df_analyze): Likewise.
31682 (loop_inverted_post_order_compute): Change argument to be a vec *.
31683 (df_analyze_loop): Adjust.
31684 (df_get_n_blocks): Likewise.
31685 (df_get_postorder): Likewise.
31686 * df.h (struct df_d): Change field to be a vec.
31687 * lcm.c (compute_laterin): Adjust.
31688 (compute_available): Likewise.
31689 * lra-lives.c (lra_create_live_ranges_1): Likewise.
31690 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
31691 * tree-ssa-pre.c (compute_antic): Likewise.
31693 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31695 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
31696 (depth_first_search::depth_first_search): Change structure init
31697 function to this constructor.
31698 (depth_first_search::add_bb): Rename function to this member.
31699 (depth_first_search::execute): Likewise.
31700 (flow_dfs_compute_reverse_finish): Adjust.
31702 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31704 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
31705 (longest_simple_path): Likewise.
31706 * shrink-wrap.c (spread_components): Likewise.
31707 (disqualify_problematic_components): Likewise.
31708 (emit_common_heads_for_components): Likewise.
31709 (emit_common_tails_for_components): Likewise.
31710 (insert_prologue_epilogue_for_components): Likewise.
31712 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31714 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
31717 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31719 * df-core.c (df_set_blocks): Start using auto_bitmap.
31720 (df_compact_blocks): Likewise.
31721 * df-problems.c (df_rd_confluence_n): Likewise.
31722 * df-scan.c (df_insn_rescan_all): Likewise.
31723 (df_process_deferred_rescans): Likewise.
31724 (df_update_entry_block_defs): Likewise.
31725 (df_update_exit_block_uses): Likewise.
31726 (df_entry_block_bitmap_verify): Likewise.
31727 (df_exit_block_bitmap_verify): Likewise.
31728 (df_scan_verify): Likewise.
31729 * lra-constraints.c (lra_constraints): Likewise.
31730 (undo_optional_reloads): Likewise.
31731 (lra_undo_inheritance): Likewise.
31732 * lra-remat.c (calculate_gen_cands): Likewise.
31733 (do_remat): Likewise.
31734 * lra-spills.c (assign_spill_hard_regs): Likewise.
31735 (spill_pseudos): Likewise.
31736 * tree-ssa-pre.c (bitmap_set_and): Likewise.
31737 (bitmap_set_subtract_values): Likewise.
31739 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31741 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
31742 management with auto_bitmap.
31743 (fix_inter_tick): Likewise.
31744 (fix_recovery_deps): Likewise.
31745 * ira.c (add_store_equivs): Likewise.
31746 (find_moveable_pseudos): Likewise.
31747 (split_live_ranges_for_shrink_wrap): Likewise.
31748 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
31749 (rtx_reuse_manager::seen_def_p): Likewise.
31750 (rtx_reuse_manager::set_seen_def): Likewise.
31751 * print-rtl.h (class rtx_reuse_manager): Likewise.
31753 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31755 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
31757 (migrate_btr_def): Likewise.
31758 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
31759 * df-core.c (loop_post_order_compute): Likewise.
31760 (loop_inverted_post_order_compute): Likewise.
31761 * hsa-common.h: Likewise.
31762 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
31763 * init-regs.c (initialize_uninitialized_regs): Likewise.
31764 * ipa-inline.c (resolve_noninline_speculation): Likewise.
31765 (inline_small_functions): Likewise.
31766 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
31767 * ira.c (combine_and_move_insns): Likewise.
31768 (build_insn_chain): Likewise.
31769 * loop-invariant.c (find_invariants): Likewise.
31770 * lower-subreg.c (propagate_pseudo_copies): Likewise.
31771 * predict.c (tree_predict_by_opcode): Likewise.
31772 (predict_paths_leading_to): Likewise.
31773 (predict_paths_leading_to_edge): Likewise.
31774 (estimate_loops_at_level): Likewise.
31775 (estimate_loops): Likewise.
31776 * shrink-wrap.c (try_shrink_wrapping): Likewise.
31777 (spread_components): Likewise.
31778 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
31779 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
31780 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
31781 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
31782 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
31783 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
31784 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
31785 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
31786 (mark_threaded_blocks): Likewise.
31787 (thread_through_all_blocks): Likewise.
31788 * tree-ssa.c (verify_ssa): Likewise.
31789 (execute_update_addresses_taken): Likewise.
31790 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
31792 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31794 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
31796 (post_order_compute): Likewise.
31797 (inverted_post_order_compute): Likewise.
31798 (pre_and_rev_post_order_compute_fn): Likewise.
31800 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31802 * genrecog.c (int_set::int_set): Explicitly construct our
31803 auto_vec base class.
31804 * vec.h (auto_vec::auto_vec): New constructor.
31806 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31808 * bitmap.h (class auto_bitmap): New constructor taking
31809 bitmap_obstack * argument.
31811 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31813 * bitmap.h (class auto_bitmap): Change type of m_bits to
31814 bitmap_head, and adjust ctor / dtor and member operators.
31816 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31818 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31819 when returned register mode doesn't match original mode.
31821 2017-05-12 Jeff Law <law@redhat.com>
31822 Jakub Jelinek <jakub@redhat.com>
31824 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31825 we look for cc setter after the compare-elim changes.
31826 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31827 within the vector to match what compare-elim now expects.
31828 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31829 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31831 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31832 after the compare-elim changes.
31833 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31834 the vector to match what compare-elim now expects.
31835 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31836 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31837 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31838 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31839 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31841 * config/visium/visium.c (single_set_and_flags): Fix where
31842 we look for cc setter after the compare-elim changes.
31843 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
31844 with the vector to match what compare-elim now expects.
31845 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
31846 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31847 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31848 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31849 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31850 (neg<mode>2_insn_set_overflow): Likewise.
31852 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31854 PR middle-end/79794
31855 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31856 maybe_expand_insn call, set ops[0].target. If still set after call,
31857 set alt_rtl. Add extra arg to recursive calls.
31858 (extract_bit_field): Add alt_rtl argument. Pass to
31860 * expmed.h (extract_bit_field): Fix prototype.
31861 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31862 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31863 to extract_bit_field_calls.
31864 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31865 Pass alt_rtl to extract_bit_field calls.
31866 * calls.c (store_unaligned_arguments_into_psuedos)
31867 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31868 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31870 * optabs.h (struct expand_operand): Add target bitfield.
31872 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31874 * compare-elim.c (try_eliminate_compare): Canonicalize
31875 operation with embedded compare to
31876 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31877 (set (reg) (operation)].
31879 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31881 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31884 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31885 cost of adding a carry flag for ADC instruction.
31886 [case MINUS]: Ignore the cost of subtracting a carry flag
31887 for SBB instruction.
31889 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31891 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31893 * config/rs6000/bmiintrin.h: New file.
31894 * config/rs6000/bmi2intrin.h: New file.
31895 * config/rs6000/x86intrin.h: New file.
31897 2017-05-12 Jeff Law <law@redhat.com>
31899 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31902 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31904 PR middle-end/80707
31905 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31907 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31909 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31910 early expansion of vector divide builtins.
31911 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
31912 builtins identified as having unsigned arguments.
31914 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31916 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
31917 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
31918 expansion of vector logical operations (and, andc, or, xor,
31921 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31923 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
31924 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
31926 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31928 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31929 early GIMPLE expansion of vector multiplies.
31931 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31933 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
31934 TARGET_HAVE_MOVT conditional.
31935 (movt splitter): Likewise.
31937 2017-05-12 Richard Biener <rguenther@suse.de>
31939 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
31940 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31941 Fold all stmts not inplace.
31943 2017-05-12 Richard Biener <rguenther@suse.de>
31945 PR tree-optimization/80713
31946 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
31947 inserted_exprs bit for not removed stmts.
31949 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
31951 PR middle-end/69921
31952 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
31953 parallelized" attribute for parallelized OpenACC kernels.
31954 * omp-offload.c (execute_oacc_device_lower): Use it.
31956 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
31957 Set "oacc kernels" attribute.
31958 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
31959 parameter. Adjust all users.
31960 (oacc_fn_attrib_kernels_p): Remove function.
31961 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
31962 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
31963 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
31964 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
31965 assert "oacc kernels" attribute is set.
31967 2017-05-11 Carl Love <cel@us.ibm.com>
31969 * config/rs6000/rs6000-c: Add support for built-in functions
31970 vector unsigned char vec_popcnt (vector signed char)
31971 vector unsigned char vec_popcnt (vector unsigned char)
31972 vector unsigned short vec_popcnt (vector signed short)
31973 vector unsigned short vec_popcnt (vector unsigned short)
31974 vector unsigned int vec_popcnt (vector signed int)
31975 vector unsigned int vec_popcnt (vector unsigned int)
31976 vector unsigned long long vec_popcnt (vector signed long long)
31977 vector unsigned long long vec_popcnt (vector unsigned long long)
31978 vector signed long long vec_slo (vector signed long long,
31979 vector signed char)
31980 vector signed long long vec_slo (vector signed long long,
31981 vector unsigned char)
31982 vector unsigned long long vec_slo (vector unsigned long long,
31983 vector signed char)
31984 vector unsigned long long vec_slo (vector unsigned long long,
31985 vector unsigned char)
31986 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
31987 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
31988 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
31989 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
31990 * doc/extend.texi: Update the built-in documentation file for the
31991 new built-in functions.
31993 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
31995 * attribs.h (sorted_attr_string): Move machine independent
31996 functions for target clone support from the i386 port to common
31997 code. Rename ix86_function_versions to common_function_versions.
31998 Rename make_name to make_unique_name.
31999 (common_function_versions): Likewise.
32000 (make_unique_name): Likewise.
32001 (make_dispatcher_decl): Likewise.
32002 (is_function_default_version): Likewise.
32003 * attribs.c (attr_strcmp): Likewise.
32004 (sorted_attr_string): Likewise.
32005 (common_function_versions): Likewise.
32006 (make_unique_name): Likewise.
32007 (make_dispatcher_decl): Likewise.
32008 (is_function_default_version): Likewise.
32009 * config/i386/i386.c (attr_strcmp): Likewise.
32010 (sorted_attr_string): Likewise.
32011 (ix86_function_versions): Likewise.
32012 (make_name): Likewise.
32013 (make_dispatcher_decl): Likewise.
32014 (is_function_default_version): Likewise.
32015 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
32017 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32020 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
32021 Account for direct move costs for vec_construct of integer
32024 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
32027 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
32028 (UNSPEC_STX_ATOMIC): Ditto.
32029 (loaddi_via_sse): New insn.
32030 (storedi_via_sse): Ditto.
32031 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
32032 Update corresponding peephole2 patterns.
32033 (atomic_storedi_fpu): Ditto.
32035 2017-05-11 Julia Koval <julia.koval@intel.com>
32037 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
32038 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
32040 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
32041 (__builtin_ia32_rsqrt14ss_mask): New builtins.
32042 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
32044 2017-05-11 Nathan Sidwell <nathan@acm.org>
32046 * graphite-poly.c: Include dumpfile.h.
32048 * dumpfle.h (dump_function): Declare here ...
32049 * tree-dump.h (dump_function): ... not here.
32050 * dumpfile.c: #include tree-cfg.h.
32051 (dump_function): Move here from ...
32052 * tree-dump.c (dump_function): ... here.
32053 * gimplify.c: #include splay-tree.h, not tree-dump.h.
32054 * graphite-poly.c: Don't include tree-dump.h.
32055 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
32056 * print-tree.c: Likewise.
32057 * stor-layout.c: Likewise.
32058 * tree-nested.c: Likewise.
32060 * dumpfile.c (dump_start): Use TDF_FLAGS.
32061 (dump_enable_all): Fix TDF_KIND check thinko.
32063 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
32065 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32066 array entries to represent two legal parameterizations of the
32067 overloaded __builtin_cmpb function, as represented by the
32068 P6_OV_BUILTIN_CMPB constant.
32069 (altivec_resolve_overloaded_builtin): Add special case handling
32070 for the __builtin_cmpb function, as represented by the
32071 P6_OV_BUILTIN_CMPB constant.
32072 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
32073 (BU_P6_64BIT_2): New macro.
32074 (BU_P6_OVERLOAD_2): New macro
32075 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
32076 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
32077 (CMPB): Add overload support to represent both 32-bit and 64-bit
32078 compare-bytes function.
32079 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
32080 support for TARGET_CMPB.
32081 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
32082 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
32083 documentation of the __builtin_cmpb overloaded built-in function.
32085 2017-05-11 Richard Biener <rguenther@suse.de>
32087 PR tree-optimization/80705
32088 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
32089 bases are not vectorizable.
32091 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32093 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
32094 when counting register pressure.
32096 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32098 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
32099 (iv_ca_more_deps): Renamed to ...
32100 (iv_ca_compare_deps): ... this.
32101 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
32103 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32105 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
32107 (determine_group_iv_costs): ... here.
32108 (find_inv_vars_cb): Record inv var if it's not recorded before.
32110 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32112 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
32113 (get_shiftadd_cost): Ditto.
32115 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32117 * tree-ssa-address.c: Include header file.
32118 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
32120 (add_to_parts): Refactor.
32121 (addr_to_parts): New parameter. Update use of move_hint_to_base.
32122 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
32125 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32127 PR tree-optimization/53090
32128 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
32130 (extract_cond_operands): Detect condition with IV on both sides
32131 and return COMP_IV_EXPR_2.
32132 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
32133 (rewrite_use_compare): Simplify by removing call to function
32134 extract_cond_operands.
32136 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32138 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
32139 (extract_cond_operands): Detect condition comparing against non-
32140 invariant bound and return appropriate enum value.
32141 (find_interesting_uses_cond): Update use of extract_cond_operands.
32142 Handle its return value accordingly.
32143 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
32145 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32147 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
32148 nonlinear iv_use computation in loop invariant sensitive way.
32150 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32152 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
32153 (find_iv_candidates): Call relate_compare_use_with_all_cands.
32155 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32157 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
32158 (dump_cand): Support iv_cand.inv_exprs.
32159 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
32161 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
32164 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32166 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
32168 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
32169 as local function. Include necessary header files.
32170 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
32172 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32174 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
32176 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32178 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
32179 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
32180 RSHIFT_EXPR and BIT_NOT_EXPR.
32182 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32184 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
32185 (adjust_setup_cost): New parameter supporting round up adjustment.
32186 (struct address_cost_data): Delete.
32187 (force_expr_to_var_cost): Don't bound cost with spill_cost.
32188 (split_address_cost, ptr_difference_cost): Delete.
32189 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
32190 (struct ainc_cost_data): New struct.
32191 (get_address_cost_ainc): New function.
32192 (get_address_cost, get_computation_cost): Reimplement.
32193 (determine_group_iv_cost_address): Record inv_expr for all uses of
32195 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
32196 (iv_ca_has_deps): Reimplemented to ...
32197 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
32199 (iv_ca_extend): Call iv_ca_more_deps.
32201 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32203 * tree-ssa-address.c (struct mem_address): Move to header file.
32204 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
32205 * tree-ssa-address.h (struct mem_address): Move from C file.
32206 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
32208 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32210 * tree-affine.h (aff_combination_type): New interface.
32211 (aff_combination_zero_p): Remove static.
32212 (aff_combination_const_p): New interface.
32213 (aff_combination_singleton_var_p): New interfaces.
32215 2017-05-11 Richard Biener <rguenther@suse.de>
32217 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
32218 Skip unreachable blocks and destinations.
32219 (eliminate): Move stmt removal and fixup ...
32220 (fini_eliminate): ... here. Skip inserted exprs.
32221 (pass_pre::execute): Move fini_pre after fini_eliminate.
32222 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
32223 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
32224 PRE to get rid of dead code that has invalid SSA form and
32225 split critical edges again.
32227 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32229 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32231 2017-05-11 Richard Biener <rguenther@suse.de>
32233 * passes.c (execute_function_todo): Verify loops if they are
32234 said to be up-to-date.
32235 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
32236 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
32238 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
32241 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
32242 handle calling assemble_external ourself.
32245 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
32246 modes with zero size. Enhance comment.
32248 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32250 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
32251 built-ins for vec_xl and vec_xst with short and char pointer
32254 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32256 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
32257 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
32258 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
32259 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
32260 (_mm_maskz_min_round_ss): New intrinsics.
32261 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
32262 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
32263 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
32264 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
32265 (__builtin_ia32_minss_mask_round): New builtins.
32266 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32267 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32268 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
32270 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
32271 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
32273 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
32276 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32278 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
32279 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
32280 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
32281 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
32282 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
32283 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
32284 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
32285 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32286 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32287 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
32288 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
32289 (__builtin_ia32_mulss_mask_round): New builtins.
32290 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32291 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32292 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
32294 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
32295 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32297 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32300 2017-05-10 Julia Koval <julia.koval@intel.com>
32302 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
32303 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
32304 (_mm256_setr_m128i): New intrinsics.
32306 2017-05-10 Julia Koval <julia.koval@intel.com>
32308 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
32309 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
32310 (_mm_maskz_rcp14_ss): New intrinsics.
32311 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
32312 (__builtin_ia32_rcp14ss_mask): New builtins.
32313 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
32315 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
32317 PR tree-optimization/51513
32318 * tree-cfg.c (gimple_seq_unreachable_p): New function.
32319 (assert_unreachable_fallthru_edge_p): Use it.
32320 (group_case_labels_stmt): Likewise.
32321 * tree-cfg.h: Prototype it.
32322 * stmt.c: Include cfghooks.h and tree-cfg.h.
32323 (emit_case_dispatch_table) <gap_label>: New local variable.
32324 Use it to fill dispatch table gaps.
32325 Test for default_label before updating probabilities.
32326 (expand_case) <default_label>: Remove unneeded initialization.
32327 Test for unreachable default case statement and remove its edge.
32328 Set default_label accordingly.
32329 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
32331 2017-05-10 Carl Love <cel@us.ibm.com>
32333 * config/rs6000/rs6000-c: Add support for built-in functions
32334 vector signed char vec_neg (vector signed char)
32335 vector signed short int vec_neg (vector short int)
32336 vector signed int vec_neg (vector signed int)
32337 vector signed long long vec_neg (vector signed long long)
32338 vector float vec_neg (vector float)
32339 vector double vec_neg (vector double)
32340 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
32342 * config/rs6000/altivec.h: Add define for vec_neg
32343 * doc/extend.texi: Update the built-in documentation for the
32344 new built-in functions.
32346 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32348 PR tree-optimization/77644
32349 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
32351 2017-05-10 Nathan Sidwell <nathan@acm.org>
32353 * dumpfile.h (TDI_lang_all): New.
32354 (TDF_KIND): New. Renumber others
32355 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
32357 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
32359 (get_dump_file_name): Adjust suffix generation.
32360 (dump_enable_all): Use TDF_KIND.
32361 * doc/invoke.texi (-fdump-lang-all): Document.
32363 * dumpfile.h: Tabify.
32365 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
32368 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
32369 Move member access before delete.
32371 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
32373 * tree-inline.c (expand_call_inline): Split block at stmt
32376 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
32379 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
32380 are now unused after splitting mov{sf,sd}_hardfloat.
32381 (f32_lr2): Likewise.
32382 (f32_lm): Likewise.
32383 (f32_lm2): Likewise.
32384 (f32_li): Likewise.
32385 (f32_li2): Likewise.
32386 (f32_lv): Likewise.
32387 (f32_sr): Likewise.
32388 (f32_sr2): Likewise.
32389 (f32_sm): Likewise.
32390 (f32_sm2): Likewise.
32391 (f32_si): Likewise.
32392 (f32_si2): Likewise.
32393 (f32_sv): Likewise.
32394 (f32_dm): Likewise.
32395 (f32_vsx): Likewise.
32396 (f32_av): Likewise.
32397 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
32398 For movsf, order stores so the VSX stores occur before the GPR
32399 store which encourages the register allocator to use a traditional
32400 FPR instead of a GPR. For movsd, order the stores so that the GPR
32401 store comes before the VSX stores to allow the power6 to work.
32402 This is due to the power6 not having a 32-bit integer store
32403 instruction from a FPR.
32404 (movsf_hardfloat): Likewise.
32405 (movsd_hardfloat): Likewise.
32407 2017-05-09 Martin Sebor <msebor@redhat.com>
32409 PR translation/80280
32410 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
32413 PR translation/80280
32414 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
32415 data member added in r247778.
32416 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
32418 2017-05-09 Nathan Sidwell <nathan@acm.org>
32420 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
32422 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
32425 2017-05-09 Marek Polacek <polacek@redhat.com>
32427 * doc/invoke.texi: Fix typo.
32429 2017-05-09 Richard Biener <rguenther@suse.de>
32431 * tree-vrp.c (vrp_val_is_max): Adjust comment.
32432 (vrp_val_is_min): Likewise.
32433 (set_value_range_to_value): Likewise.
32434 (set_value_range_to_nonnegative): Likewise.
32435 (gimple_assign_nonzero_p): Likewise.
32436 (gimple_stmt_nonzero_p): Likewise.
32437 (vrp_int_const_binop): Likewise. Remove unreachable case.
32438 (adjust_range_with_scev): Adjust comments.
32439 (compare_range_with_value): Likewise.
32440 (extract_range_from_phi_node): Likewise.
32441 (test_for_singularity): Likewise.
32443 2017-05-09 Richard Biener <rguenther@suse.de>
32445 * tree-vrp.c (get_single_symbol): Add assert that we don't
32446 get overflowed constants as invariant part.
32447 (compare_values_warnv): Add comment before the TREE_NO_WARNING
32448 checks. Use wi::cmp instead of recursing for integer constants.
32449 (compare_values): Just ignore whether we assumed undefined
32450 overflow instead of failing the compare.
32451 (extract_range_for_var_from_comparison_expr): Add comment before the
32452 TREE_NO_WARNING sets.
32453 (test_for_singularity): Likewise.
32454 (extract_range_from_comparison): Do not disable optimization
32455 when we assumed undefined overflow.
32456 (extract_range_basic): Remove init of unused var.
32458 2017-05-09 Richard Biener <rguenther@suse.de>
32460 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
32461 (extract_range_from_multiplicative_op_1): Adjust.
32462 (extract_range_from_binary_expr_1): Use int_const_binop.
32464 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
32467 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
32468 rs6000_store_data_bypass_p in seven define_bypass directives and
32469 in several comments.
32470 * config/rs6000/rs6000-protos.h: Add prototype for
32471 rs6000_store_data_bypass_p function.
32472 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
32473 function implements slightly different (rs6000-specific) semantics
32474 than store_data_bypass_p, returning false rather than aborting
32475 with assertion error when arguments do not satisfy the
32476 requirements of store data bypass.
32477 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
32478 rs6000_store_data_bypass_p.
32480 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
32482 * config/xtensa/xtensa-protos.h
32483 (xtensa_initial_elimination_offset): New declaration.
32484 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
32485 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
32486 macro definition, add case for FRAME_POINTER_REGNUM when
32487 FRAME_GROWS_DOWNWARD.
32488 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
32489 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
32490 xtensa_initial_elimination_offset.
32492 2017-05-08 Nathan Sidwell <nathan@acm.org>
32494 * doc/invoke.texi: Alphabetize -fdump options.
32496 2017-05-08 Martin Sebor <msebor@redhat.com>
32498 PR translation/80280
32499 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
32501 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
32503 * target.def (compute_frame_layout): New optional target hook.
32504 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
32505 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
32506 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
32508 * reload1.c (verify_initial_elim_offsets): Likewise.
32509 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
32510 (use_simple_return_p): Call arm_compute_frame_layout if needed.
32511 (arm_get_frame_offsets): Split up into this ...
32512 (arm_compute_frame_layout): ... and this function.
32514 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
32516 * config/aarch64/constraints.md (Usa): New constraint.
32517 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
32519 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
32521 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
32522 with_multilib_list after it has been checked.
32524 2017-05-08 Richard Biener <rguenther@suse.de>
32526 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
32527 (bitmap_set_subtract_values): Likewise.
32529 2017-05-08 Richard Biener <rguenther@suse.de>
32531 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
32532 (gimple_assign_nonzero): ... this and remove strict_overflow_p
32534 (gimple_stmt_nonzero_warnv_p): Rename to ...
32535 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
32537 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
32538 (extract_range_basic): Adjust, do not disable propagation on
32539 strict overflow sensitive simplification.
32540 (vrp_visit_cond_stmt): Likewise.
32542 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
32544 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
32545 body size unconditionally.
32547 2017-05-07 Jeff Law <law@redhat.com>
32550 2017-05-06 Jeff Law <law@redhat.com>
32551 PR tree-optimization/78496
32552 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32555 PR tree-optimization/78496
32556 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32557 (simplify_stmt_using_ranges): Call it.
32558 (vrp_dom_walker::before_dom_children): Extract equivalences
32559 from an ASSERT_EXPR with an equality comparison against a
32562 2017-05-06 Jeff Law <law@redhat.com>
32564 PR tree-optimization/78496
32565 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32568 PR tree-optimization/78496
32569 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32570 (simplify_stmt_using_ranges): Call it.
32571 (vrp_dom_walker::before_dom_children): Extract equivalences
32572 from an ASSERT_EXPR with an equality comparison against a
32575 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32577 * lra-constraints.c (lra_copy_reg_equiv): New function.
32578 (split_reg): Use it to copy equivalence information from the
32579 original register to the spill register.
32581 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32583 PR rtl-optimization/75964
32584 * simplify-rtx.c (simplify_const_relational_operation): Remove
32585 invalid handling of comparisons of integer ABS.
32587 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
32589 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
32590 initialize to zero.
32591 (init_regs): Remove declaration.
32592 (function_arg_advance_32): Initialize error_p as boolean variable.
32594 2017-05-05 Nathan Sidwell <nathan@acm.org>
32596 * store-motion.c (remove_reachable_equiv_notes): Reformat long
32597 lines. Use for (;;).
32599 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32601 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
32602 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
32603 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
32604 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
32605 VF=2 that require versioning.
32607 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32609 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
32612 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32614 * diagnostic.h (diagnostic_override_option_index): Convert from
32615 macro to inline function.
32617 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32619 * diagnostic.c (last_module_changed_p): New function.
32620 (set_last_module): New function.
32621 (diagnostic_report_current_module): Convert macro usage to
32622 the above functions.
32623 * diagnostic.h (diagnostic_context::last_module): Strengthen
32624 from const line_map * to const line_map_ordinary *.
32625 (diagnostic_last_module_changed): Delete macro.
32626 (diagnostic_set_last_module): Delete macro.
32628 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32630 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
32631 with diagnostic_report_diagnostic.
32632 (diagnostic_n_impl_richloc): Likewise.
32633 * diagnostic.h (report_diagnostic): Delete macro.
32634 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
32635 with diagnostic_report_diagnostic.
32636 * substring-locations.c (format_warning_va): Likewise.
32638 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32640 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
32641 save/restor of format_spec. Move option-printing code to...
32642 (print_option_information): ...this new function, and
32643 reimplement by simply printing to the pretty_printer,
32644 rather than appending to the format string.
32646 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32648 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
32649 handling logic into...
32650 (update_effective_level_from_pragmas): ...this new function.
32652 2017-05-04 Andrew Waterman <andrew@sifive.com>
32654 * config/riscv/riscv.opt (mstrict-align): New option.
32655 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
32656 (SLOW_UNALIGNED_ACCESS): Define.
32657 (riscv_slow_unaligned_access): Declare.
32658 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
32660 (riscv_slow_unaligned_access): New variable.
32661 (rocket_tune_info): Set slow_unaligned_access to true.
32662 (optimize_size_tune_info): Set slow_unaligned_access to false.
32663 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
32664 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
32665 (riscv_option_override): Set riscv_slow_unaligned_access.
32666 * doc/invoke.texi: Add -mstrict-align to RISC-V.
32668 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
32670 * config/riscv/riscv.md: Unify indentation.
32672 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
32677 * config/rs6000/rs6000.md (u code attribute): Add FIX and
32679 (extendsi<mode>2): Add support for doing sign extension via
32680 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
32681 don't have ISA 3.0 instructions.
32682 (extendsi<mode>2 splitter): Likewise.
32683 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
32684 generate the normal insns since SImode can now go in vector
32685 registers. Disallow the special UNSPECs needed for previous
32686 machines to hide SImode being used. Add new insns
32687 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
32688 (fix_trunc<mode>si2_stfiwx): Likewise.
32689 (fix_trunc<mode>si2_internal): Likewise.
32690 (fixuns_trunc<mode>si2): Likewise.
32691 (fixuns_trunc<mode>si2_stfiwx): Likewise.
32692 (fctiw<u>z_<mode>_smallint): Likewise.
32693 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
32694 of floating point to 32-bit integer from doing a direct move to
32695 the GPR registers to do a store.
32696 (fctiwz_<mode>): Break long line.
32698 2017-05-05 Bin Cheng <bin.cheng@arm.com>
32700 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
32701 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
32702 (addr_list, addr_offset_valid_p): New.
32703 (split_address_groups): Check offset validity with above function.
32704 (gt-tree-ssa-loop-ivopts.h): Include header file.
32706 2017-05-05 Nathan Sidwell <nathan@acm.org>
32708 * config.gcc (arm*-*-*): Add missing 'fi'.
32710 2017-05-05 Steve Ellcey <sellcey@cavium.com>
32712 * doc/invoke.texi (-fopt-info): Explicitly say order of options
32713 included in -fopt-info does not matter.
32714 * doc/optinfo.texi (-fopt-info): Fix description of default
32715 behavour. Explicitly say order of options included in -fopt-info
32718 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
32720 * config.gcc: Allow combinations of aprofile and rmprofile values for
32721 --with-multilib-list.
32722 * config/arm/t-multilib: New file.
32723 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
32724 variables. Remove setting of ISA and floating-point ABI in
32725 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
32726 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
32727 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
32728 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
32730 * config/arm/t-rmprofile: Likewise except for the matches changes.
32731 * doc/install.texi (--with-multilib-list): Document the combination of
32732 aprofile and rmprofile values and warn about pitfalls in doing that.
32734 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32736 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
32737 (movdi_aarch64): Likewise.
32739 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32741 PR tree-optimization/80632
32742 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
32744 (build_arrays): Initialize it for virtual phis.
32745 (fix_phi_nodes): Use it for virtual phis.
32747 PR tree-optimization/80558
32748 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
32749 [x, y] op z into [x op, y op z] for op & or | if conditions
32752 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
32753 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32756 * config/arm/arm.md (use_literal_pool): Remove.
32757 (64-bit immediate split): No longer takes cost into consideration
32758 if arm_disable_literal_pool is enabled.
32759 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
32760 used when arm_disable_literal_pool is enabled.
32761 (arm_max_const_double_inline_cost): Remove use of
32762 arm_disable_literal_pool.
32763 (push_minipool_fix): Add assert.
32764 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
32765 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
32766 (no_literal_pool_sf_immediate): New.
32768 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32770 PR tree-optimization/80613
32771 * tree-ssa-dce.c (propagate_necessity): Remove cases for
32772 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
32774 2017-05-05 Richard Biener <rguenther@suse.de>
32776 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
32778 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
32780 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
32781 of this flag from insn conditions due to removal from r247495.
32783 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32785 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
32787 (arm_early_store_addr_dep_ptr): Likewise.
32788 * config/arm/aarch-common-protos.h
32789 (arm_early_load_addr_dep_ptr): Add prototype.
32790 (arm_early_store_addr_dep_ptr): Likewise.
32791 * config/arm/cortex-a53.md: Add new bypasses.
32793 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32795 * tree.c (next_type_uid): Change type to unsigned.
32796 (type_hash_canon): Decrement back next_type_uid if
32797 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
32798 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
32801 2017-05-04 Martin Sebor <msebor@redhat.com>
32803 * builtins.c: Fix a trivial typo in a comment.
32805 PR middle-end/79234
32806 * builtins.c (check_sizes): Adjust to handle reading past the end.
32807 Avoid printing excessive upper bound of ranges. Use %E to print
32808 tree nodes instead of converting them to %wu.
32809 (expand_builtin_memchr): New function.
32810 (compute_dest_size): Rename...
32811 (compute_objsize): ...to this.
32812 (expand_builtin_memcpy): Adjust.
32813 (expand_builtin_mempcpy): Adjust.
32814 (expand_builtin_strcat): Adjust.
32815 (expand_builtin_strcpy): Adjust.
32816 (check_strncat_sizes): Adjust.
32817 (expand_builtin_strncat): Adjust.
32818 (expand_builtin_strncpy): Adjust and simplify.
32819 (expand_builtin_memset): Adjust.
32820 (expand_builtin_bzero): Adjust.
32821 (expand_builtin_memcmp): Adjust.
32822 (expand_builtin): Handle memcmp.
32823 (maybe_emit_chk_warning): Check strncat just once.
32825 2017-05-04 Martin Sebor <msebor@redhat.com>
32827 PR preprocessor/79214
32828 PR middle-end/79222
32829 PR middle-end/79223
32830 * builtins.c (check_sizes): Add inlining context and issue
32831 warnings even when -Wno-system-headers is set.
32832 (check_strncat_sizes): Same.
32833 (expand_builtin_strncat): Same.
32834 (expand_builtin_memmove): New function.
32835 (expand_builtin_stpncpy): Same.
32836 (expand_builtin): Handle memmove and stpncpy.
32838 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32840 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32841 which is not used any more.
32843 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32845 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32847 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32849 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32850 (cortexa53_tunings): Likewise.
32851 (cortexa57_tunings): Likewise.
32852 (cortexa72_tunings): Likewise.
32853 (cortexa73_tunings): Likewise.
32855 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32857 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32858 Set loop alignment to 8.
32860 2017-05-04 Martin Sebor <msebor@redhat.com>
32862 PR translation/80280
32863 * builtins.c (expand_builtin_object_size): Add missing quoting to
32864 %D and like directives.
32865 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32866 (hsa_type_for_tree_type): Same.
32867 (verify_function_arguments): Same.
32868 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32869 * varasm.c (get_section): Same.
32872 2017-05-04 Martin Sebor <msebor@redhat.com>
32874 PR translation/80280
32875 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32877 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32879 * config/aarch64/aarch64.c (generic_addrcost_table):
32880 Change HI/TI mode setting.
32882 2017-05-04 Martin Jambor <mjambor@suse.cz>
32884 PR tree-optimization/80622
32885 * tree-sra.c (comes_initialized_p): New function.
32886 (build_accesses_from_assign): Only set write lazily when
32887 comes_initialized_p is false.
32888 (analyze_access_subtree): Use comes_initialized_p.
32889 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32890 instead of testing for PARM_DECL.
32892 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32894 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32895 constraint on operand 0 to allow more general addressing modes.
32896 Adjust output template.
32897 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32899 * config/aarch64/aarch64-protos.h
32900 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32901 * config/aarch64/constraints.md (Dp): New address constraint.
32902 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32905 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32907 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32908 update use of estimate_ipcp_clone_size_and_time.
32909 (estimate_local_effects): Update use of
32910 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32911 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
32912 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
32913 Return nonspecialized time.
32915 2017-05-04 Richard Biener <rguenther@suse.de>
32917 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
32918 for the last VUSE which def dominates the PHI. Directly call
32920 (get_continuation_for_phi_1): Remove.
32922 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
32924 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
32925 to explain the use of truncating division. Cap the number of
32926 iterations to the maximum given by nb_iterations_upper_bound,
32929 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32931 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
32932 * configure: Regenerate.
32933 * config.in: Regenerate.
32934 * config/i386/driver-mingw32.c: new file.
32935 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
32936 * config.host: Link driver-mingw32.o on MinGW host.
32937 * doc/install.texi: Document new --enable-mingw-wildcard configure
32940 2017-05-04 Marek Polacek <polacek@redhat.com>
32942 PR tree-optimization/80612
32943 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
32945 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32946 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
32948 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
32949 (movt splitter): Likewise.
32950 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
32951 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
32952 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
32953 block for Thumb-1 with MOVT.
32954 (thumb2_legitimate_address_p): Move code block ...
32955 (can_avoid_literal_pool_for_label_p): ... into this new function.
32956 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
32958 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
32959 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
32960 "M-profile targets with the MOVT instruction".
32962 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32964 * config/arm/arm-builtins.c (arm_init_builtins): Rename
32965 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
32966 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
32968 2017-05-04 Martin Liska <mliska@suse.cz>
32970 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
32971 variable cond_code.
32973 2017-05-04 Richard Biener <rguenther@suse.de>
32975 * tree.c (array_at_struct_end_p): Handle arrays at struct
32976 end with flexarrays more conservatively. Refactor and treat
32977 arrays of arrays or aggregates more strict. Fix
32978 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
32979 * tree.c (array_at_struct_end_p): Adjust prototype.
32980 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
32981 * gimple-fold.c (get_range_strlen): Likewise.
32982 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
32984 2017-05-04 Richard Biener <rguenther@suse.de>
32986 PR tree-optimization/31130
32987 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
32989 (supports_overflow_infinity): Likewise.
32990 (is_negative_overflow_infinity): Likewise.
32991 (is_positive_overflow_infinity): Likewise.
32992 (is_overflow_infinity): Likewise.
32993 (stmt_overflow_infinity): Likewise.
32994 (overflow_infinity_range_p): Likewise.
32995 (usable_range_p): Remove as always returning true.
32996 (make_overflow_infinity): Remove.
32997 (negative_overflow_infinity): Likewise.
32998 (positive_overflow_infinity): Likewise.
32999 (avoid_overflow_infinity): Likewise.
33000 (set_value_range): Adjust accordingly.
33001 (set_value_range_to_nonnegative): Likewise, remove now unused
33002 overflow_infinity arg.
33003 (vrp_operand_equal_p): Adjust.
33004 (update_value_range): Likewise.
33005 (range_int_cst_singleton_p): Likewise.
33006 (operand_less_p): Likewise.
33007 (compare_values_warnv): Likewise.
33008 (extract_range_for_var_from_comparison_expr): Likewise.
33009 (vrp_int_const_binop): Likewise.
33010 (zero_nonzero_bits_from_vr): Likewise.
33011 (extract_range_from_multiplicative_op_1): Likewise.
33012 (extract_range_from_binary_expr_1): Likewise.
33013 (extract_range_from_unary_expr): Likewise.
33014 (extract_range_from_comparison): Likewise.
33015 (extract_range_basic): Likewise.
33016 (adjust_range_with_scev): Likewise.
33017 (compare_ranges): Likewise.
33018 (compare_range_with_value): Likewise.
33019 (dump_value_range): Likewise.
33020 (test_for_singularity): Likewise, remove strict_overflow_p parameter
33022 (simplify_cond_using_ranges): Adjust.
33024 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
33026 * brig-builtins.def: Added a builtin for class_f64.
33027 * builtin-types.def: Added a builtin type needed by class_f64.
33029 2017-05-03 Jason Merrill <jason@redhat.com>
33031 * timevar.def: Add TV_CONSTEXPR.
33033 2017-05-03 David Malcolm <dmalcolm@redhat.com>
33035 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
33037 2017-05-03 Martin Jambor <mjambor@suse.cz>
33039 * ipa-prop.c (ipa_update_after_lto_read): Removed.
33040 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
33041 * ipa-cp.c (ipcp_propagate_stage): Do not call
33042 ipa_update_after_lto_read.
33043 * ipa-inline.c (ipa_inline): Likewise.
33045 2017-05-03 Martin Jambor <mjambor@suse.cz>
33047 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
33048 tag. Added a default constructor and a destructor.
33049 (ipa_edge_args_sum_t): New class;
33050 (ipa_edge_args_sum): Declare.
33051 (ipa_edge_args_vector): Remove declaration.
33052 (IPA_EDGE_REF): Use ipa_edge_args_sum.
33053 (ipa_free_edge_args_substructures): Remove declaration.
33054 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
33055 (ipa_edge_args_info_available_for_edge_p): Likewise.
33056 * ipa-prop.c (ipa_edge_args_vector): Removed.
33057 (edge_removal_hook_holder): Likewise.
33058 (edge_duplication_hook_holder): Likewise.
33059 (ipa_edge_args_sum): New variable.
33060 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
33061 ipa_edge_args_vector.
33062 (ipa_free_edge_args_substructures): Likewise.
33063 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
33064 ipa_edge_args_vector.
33065 (ipa_edge_removal_hook): Turned into method
33066 ipa_edge_args_sum_t::remove.
33067 (ipa_edge_duplication_hook): Turned into method
33068 ipa_edge_args_sum_t::duplicate.
33069 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
33070 registering edge hooks.
33071 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
33072 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
33073 ipa_edge_args_sum instead of ipa_edge_args_vector.
33074 * ipa-profile.c (ipa_profile): Likewise.
33076 2017-05-03 Martin Jambor <mjambor@suse.cz>
33078 * symbol-summary.h (function_summary): New method exists.
33079 (function_summary::symtab_removal): Deallocate through release.
33080 (call_summary): New class.
33081 (gt_ggc_mx): New overload.
33082 (gt_pch_nx): Likewise.
33083 (gt_pch_nx): Likewise.
33085 2017-05-03 Jeff Law <law@redhat.com>
33087 PR tree-optimization/78496
33088 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
33089 from simplify_cond_using_ranges. Split off code to walk
33090 backwards through casts into ...
33091 (simplify_cond_using_ranges_2): New function.
33092 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
33093 (execute_vrp): After identifying jump threads, call
33094 simplify_cond_using_ranges_2.
33096 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
33099 * ipa-inline.h (inline_summary): Add ctor.
33100 (create_ggc): Do not use ggc_cleared_alloc.
33102 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
33103 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33105 * gcc.c (handle_braces): Support escaping in switch matching
33107 * doc/invoke.texi (Spec Files): Document it.
33108 Remove superfluous @code markup in items.
33110 2017-05-03 David Malcolm <dmalcolm@redhat.com>
33112 * diagnostic-show-locus.c (struct column_range): New struct.
33113 (get_affected_columns): New function.
33114 (get_printed_columns): New function.
33115 (struct correction): New struct.
33116 (correction::ensure_capacity): New function.
33117 (correction::ensure_terminated): New function.
33118 (struct line_corrections): New struct.
33119 (line_corrections::~line_corrections): New dtor.
33120 (line_corrections::add_hint): New function.
33121 (layout::print_trailing_fixits): Reimplement in terms of the new
33123 (selftest::test_overlapped_fixit_printing): New function.
33124 (selftest::diagnostic_show_locus_c_tests): Call it.
33126 2017-05-03 Nathan Sidwell <nathan@acm.org>
33128 Canonicalize canonical type hashing
33129 * tree.h (type_hash_canon_hash): Declare.
33130 * tree.c (type_hash_list, attribute_hash_list): Move into
33131 type_hash_canon_hash.
33132 (build_type_attribute_qual_variant): Break out hash code calc into
33133 type_hash_canon_hash.
33134 (type_hash_canon_hash): New. Generic type hash computation.
33135 (build_range_type_1, build_array_type_1, build_function_type,
33136 build_method_type_directly, build_offset_type, build_complex_type,
33137 make_vector_type): Call it.
33139 2017-05-03 Richard Biener <rguenther@suse.de>
33141 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
33142 When all DRs have unknown misaligned do not always peel
33143 when there is a store but apply the same costing model as if
33144 there were only loads.
33146 2017-05-03 Richard Biener <rguenther@suse.de>
33149 PR tree-optimization/80492
33150 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33151 compare_base_decls returning dont-know properly.
33153 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
33155 * config/arm/iterators.md (CCSI): New mode iterator.
33156 (arch): New mode attribute.
33157 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
33158 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
33159 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
33160 code iterator for success result mode.
33161 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
33162 the corresponding new insn generators.
33164 2017-05-03 Bin Cheng <bin.cheng@arm.com>
33167 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33168 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33170 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
33172 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
33173 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
33174 (DDR_A): Wrap DDR argument in brackets.
33175 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
33176 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
33177 (DDR_REVERSED_P): Likewise.
33179 2017-05-03 Jakub Jelinek <jakub@redhat.com>
33181 PR tree-optimization/79472
33182 * tree-switch-conversion.c (struct switch_conv_info): Add
33183 contiguous_range and default_case_nonstandard fields.
33184 (collect_switch_conv_info): Compute contiguous_range and
33185 default_case_nonstandard fields, don't clear final_bb if
33186 contiguous_range and only the default case doesn't have the required
33188 (check_all_empty_except_final): Set default_case_nonstandard instead
33189 of failing if contiguous_range and the default case doesn't have empty
33191 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
33192 and only the default case doesn't have the required constants. Skip
33194 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
33195 if default_case_nonstandard.
33196 (build_constructors): Build constant 1 just once. Assert that default
33197 values aren't inserted in between cases if contiguous_range. Skip
33199 (build_arrays): Skip virtual phis.
33200 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
33201 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
33202 Handle virtual phis.
33203 (gen_inbound_check): Handle default_case_nonstandard case.
33204 (process_switch): Adjust check_final_bb caller. Call
33205 gather_default_values with the first non-default case instead of
33206 default case if default_case_nonstandard.
33208 2017-05-02 Nathan Sidwell <nathan@acm.org>
33210 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
33211 check. Fix formatting.
33213 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
33215 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
33216 errors when comparing specialized and unspecialized times.
33218 2017-05-02 David Malcolm <dmalcolm@redhat.com>
33220 * diagnostic-show-locus.c
33221 (layout::should_print_annotation_line_p): Make private.
33222 (layout::print_annotation_line): Make private.
33223 (layout::annotation_line_showed_range_p): Make private.
33224 (layout::show_ruler): Make private.
33225 (layout::print_source_line): Make private. Pass in line and
33226 line_width, rather than calling location_get_source_line. Drop
33228 (layout::print_leading_fixits): New method.
33229 (layout::print_any_fixits): Rename to...
33230 (layout::print_trailing_fixits): ...this, and make private.
33231 Don't print newline fixits.
33232 (diagnostic_show_locus): Move logic for printing one row into...
33233 (layout::print_line): ...this new function. Move the
33234 location_get_source_line call and error-handling from
33235 print_source_line to here. Call print_leading_fixits, and rename
33236 print_any_fixits to print_trailing_fixits.
33237 (selftest::test_fixit_insert_containing_newline): Update now that
33238 newlines are partially supported.
33239 (selftest::test_fixit_insert_containing_newline_2): New test.
33240 (selftest::test_fixit_replace_containing_newline): Update comments.
33241 (selftest::diagnostic_show_locus_c_tests): Call the new test.
33242 * edit-context.c (class added_line): New class.
33243 (class edited_line): Describe newline handling in comment.
33244 (edited_line::actually_edited_p): New method.
33245 (edited_line::print_content): Delete redundant decl.
33246 (edited_line::m_predecessors): New field.
33247 (edited_file::print_content): Call edited_line::print_content.
33248 (edited_file::print_diff): Update to support newlines.
33249 (edited_file::print_diff_hunk): Likewise.
33250 (edited_file::print_run_of_changed_lines): New function.
33251 (edited_file::print_diff_line): Convert to...
33252 (print_diff_line): ...this.
33253 (edited_file::get_effective_line_count): New function.
33254 (edited_line::edited_line): Initialize new field m_predecessors.
33255 (edited_line::~edited_line): Clean up m_predecessors.
33256 (edited_line::apply_fixit): Handle newlines.
33257 (edited_line::get_effective_line_count): New function.
33258 (edited_line::print_content): New function.
33259 (edited_line::print_diff_lines): New function.
33260 (selftest::test_applying_fixits_insert_containing_newline): New
33262 (selftest::test_applying_fixits_replace_containing_newline): New
33264 (selftest::insert_line): New function.
33265 (selftest::test_applying_fixits_multiple_lines): Add example of
33267 (selftest::edit_context_c_tests): Call the new tests.
33269 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33271 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
33272 parameter cand. Update dump information.
33273 (get_computation_cost): Update uses.
33275 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33277 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
33278 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
33279 (get_computation_at, rewrite_use_address): Update use of
33280 get_computation_aff.
33282 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33284 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
33285 (get_computation): Delete.
33286 (get_computation_cost): Implement like get_computation_cost_at.
33287 Use get_computation_at.
33288 (get_computation_cost_at): Delete.
33289 (rewrite_use_nonlinear_expr): Use get_computation_at.
33290 (rewrite_use_compare, remove_unused_ivs): Ditto.
33292 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33294 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
33296 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33298 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
33299 (ivopts_global_cost_for_size): Rename parameter and update uses.
33300 (iv_ca_recount_cost): Update uses.
33301 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
33302 candidates seperately in n_invs and n_cands.
33303 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
33305 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33307 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
33308 (find_inv_vars_cb): New.
33309 (find_depends): Renamed to ...
33310 (find_inv_vars): ... this.
33311 (add_candidate_1, force_var_cost): Call find_inv_vars.
33312 (split_address_cost, determine_group_iv_cost_cond): Ditto.
33314 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33316 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
33317 inv_vars. Add inv_exprs.
33318 (struct iv_cand): Rename depends_on to inv_vars.
33319 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
33320 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
33321 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
33322 (dump_cand): Dump inv_vars.
33323 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
33324 (record_invariant, find_depends, add_candidate_1): Ditto.
33325 (set_group_iv_cost, force_var_cost): Ditto.
33326 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
33327 (get_computation_cost_at, get_computation_cost): Ditto.
33328 (determine_group_iv_cost_generic): Ditto.
33329 (determine_group_iv_cost_address): Ditto.
33330 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
33331 (determine_group_iv_costs): Ditto.
33332 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
33333 (iv_ca_set_remove_invariants): Renamed to ...
33334 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
33335 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
33336 (iv_ca_set_add_invariants): Renamed to ...
33337 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
33338 (iv_ca_set_cp): Use iv_ca_set_add_invs.
33339 (iv_ca_has_deps): Support inv_vars and inv_exprs.
33340 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
33341 (create_new_ivs): Remove useless dump.
33343 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33345 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
33347 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
33348 (iv_ca_set_no_cp, create_new_iv): Ditto.
33350 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33352 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33354 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33356 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
33357 function tree_check2.
33359 2017-05-02 Martin Liska <mliska@suse.cz>
33361 * doc/gcov.texi: Add missing preposition.
33362 * gcov.c (function_info::function_info): Properly fill up
33363 all member variables.
33365 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33367 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
33369 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33371 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
33373 2017-05-02 Martin Liska <mliska@suse.cz>
33376 * lto-streamer-in.c (lto_read_tree_1): Remove
33377 LTO_STREAMER_DEBUG.
33378 * lto-streamer.c (struct tree_hash_entry): Likewise.
33379 (struct tree_entry_hasher): Likewise.
33380 (tree_entry_hasher::hash): Likewise.
33381 (tree_entry_hasher::equal): Likewise.
33382 (lto_streamer_init): Likewise.
33383 (lto_orig_address_map): Likewise.
33384 (lto_orig_address_get): Likewise.
33385 (lto_orig_address_remove): Likewise.
33386 * lto-streamer.h: Likewise.
33387 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
33388 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
33390 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
33392 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
33393 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
33394 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
33395 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
33396 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
33397 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
33398 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
33399 (mm_maskz_sub_ss): New intrinsics.
33400 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33401 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
33402 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
33403 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
33404 (__builtin_ia32_subss_mask_round): New builtins.
33405 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33406 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
33407 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
33409 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
33410 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
33412 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
33415 2017-05-02 Martin Jambor <mjambor@suse.cz>
33417 PR tree-optimization/78687
33418 * tree-sra.c (access): New field parent.
33419 (process_subtree_disqualification): New function.
33420 (disqualify_candidate): Call it.
33421 (build_accesses_from_assign): Reset write flag if creating an
33423 (build_access_subtree): Fill in parent field and also prpagate
33424 down grp_write flag.
33425 (create_artificial_child_access): New parameter set_grp_write, set
33426 grp_write to its value.
33427 (propagate_subaccesses_across_link): Also propagate grp_write flag
33429 (propagate_all_subaccesses): Push the closest parent back to work
33430 queue if add_access_to_work_queue returned true.
33432 2017-05-02 Richard Biener <rguenther@suse.de>
33434 * common.opt (fstrict-overflow): Alias negative to fwrapv.
33435 * doc/invoke.texi (fstrict-overflow): Remove all traces of
33436 -fstrict-overflow documentation.
33437 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
33438 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
33439 flag_strict_overflow.
33440 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
33441 * lto-opts.c (lto_write_options): Do not stream it.
33442 * lto-wrapper.c (merge_and_complain): Do not handle it.
33443 * opts.c (default_options_table): Do not set -fstrict-overflow.
33444 (finish_options): Likewise do not clear it when sanitizing.
33445 * simplify-rtx.c (simplify_const_relational_operation): Do not
33446 test flag_strict_overflow.
33448 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33450 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
33451 using enabled attribute.
33452 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
33453 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
33454 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
33455 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
33456 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
33457 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
33458 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
33459 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
33460 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
33461 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
33463 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33465 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
33467 2017-05-02 Richard Biener <rguenther@suse.de>
33469 PR tree-optimization/80591
33471 2017-04-10 Richard Biener <rguenther@suse.de>
33473 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33476 2017-05-02 Richard Biener <rguenther@suse.de>
33478 PR tree-optimization/80549
33479 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
33480 (cleanup_tree_cfg_noloop): Create forwarders to known loop
33481 headers if they do not have a preheader.
33483 2017-05-02 Martin Liska <mliska@suse.cz>
33486 * common.opt: Fix typo.
33487 * doc/invoke.texi: Likewise.
33489 2017-05-01 Jan Beulich <jbeulich@suse.com>
33491 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
33492 swapping, add (x,x,m,x,n) alternative.
33494 2017-05-01 Nathan Sidwell <nathan@acm.org>
33496 * calls.c (combine_pending_stack_adjustment_and_call): Remove
33497 unnecessary unadjusted_alignment check.
33499 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33502 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
33504 * gimplify.c (gimplify_cilk_detach): New function.
33505 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
33506 * tree-core.h: Document EXPR_CILK_SPAWN.
33507 * tree.h (EXPR_CILK_SPAWN): Define.
33509 2017-05-01 David Malcolm <dmalcolm@redhat.com>
33511 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
33512 to use new fixit_hint representation, using the "replace" logic.
33513 (get_line_span_for_fixit_hint): Likewise.
33514 (layout::print_any_fixits): Likewise.
33515 (selftest::test_one_liner_many_fixits): Rename to...
33516 (selftest::test_one_liner_many_fixits_1): ...this, and update
33517 comment and expected output to reflect that the multiple fix-it
33518 hints are now consolidated into one insertion.
33519 (selftest::test_one_liner_many_fixits_2): New test.
33520 (selftest::test_diagnostic_show_locus_one_liner): Update for
33522 (selftest::test_fixit_consolidation): Update for fix-it API
33524 * diagnostic.c (print_parseable_fixits): Likewise.
33525 * edit-context.c (edited_line::m_line_events): Convert from
33526 auto_vec <line_event *> to auto_vec <line_event>.
33527 (class line_event): Convert from abstract base class to a concrete
33528 class, taking over the role of replace_event.
33529 (class insert_event): Delete.
33530 (class replace_event): Rename to class line_event. Convert to
33532 (edit_context::add_fixits): Reimplement.
33533 (edit_context::apply_insert): Delete.
33534 (edit_context::apply_replace): Rename to...
33535 (edit_context::apply_fixit): ...this. Convert to half-open range.
33536 (edited_file::apply_insert): Delete.
33537 (edited_file::apply_replace): Rename to...
33538 (edited_file::apply_fixit): ...this.
33539 (edited_line::~edited_line): Drop deletion of events.
33540 (edited_line::apply_insert): Delete.
33541 (edited_line::apply_replace): Rename to...
33542 (edited_line::apply_fixit): ...this. Convert to half-open range.
33543 Update for change to type of m_line_events.
33544 * edit-context.h (edit_context::apply_insert): Delete.
33545 (edit_context::apply_replace): Rename to...
33546 (edit_context::apply_fixit): ...this.
33548 2017-05-01 Martin Sebor <msebor@redhat.com>
33550 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
33553 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
33556 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
33557 __get_cpuid_max returns 0.
33558 (__get_cpuid_count): Ditto.
33560 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
33562 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
33563 replacement expression is another instance of one of its arguments.
33565 2017-05-01 Jakub Jelinek <jakub@redhat.com>
33568 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
33569 check for stack push/pop autoinc.
33570 * config/i386/i386.c (ix86_agi_dependent): Return false
33571 if the only reason why modified_in_p returned true is that
33572 addr is SP based and set_insn is a push or pop.
33574 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33576 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
33579 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33582 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
33583 (account_size_time): Use two predicates - exec_pred and
33585 (evaluate_conditions_for_known_args): Compute both clause and
33587 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
33588 (inline_summary_t::duplicate): Update.
33589 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
33591 (compute_inline_parameters): Likewise.
33592 (estimate_edge_size_and_time): Update caluclation of time.
33593 (estimate_node_size_and_time): Compute both time and nonspecialized
33595 (estimate_ipcp_clone_size_and_time): Update.
33596 (inline_merge_summary): Update.
33597 (do_estimate_edge_time): Update.
33598 (do_estimate_edge_size): Update.
33599 (do_estimate_edge_hints): Update.
33600 (inline_read_section, inline_write_summary): Stream both new predicates.
33601 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
33603 (compute_inlined_call_time): Cleanup.
33604 (big_speedup_p): Update.
33605 (edge_badness): Update.
33606 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
33607 (size_time_entry): Replace predicate by exec_predicate and
33608 nonconst_predicate.
33609 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
33610 (estimate_edge_time): Return also nonspec_time.
33611 (reset_edge_growth_cache): Update.
33613 2017-04-29 Jakub Jelinek <jakub@redhat.com>
33615 PR rtl-optimization/80491
33616 * ifcvt.c (noce_process_if_block): When looking for x setter
33617 with missing else_bb, don't check only the insn right before
33618 cond_earliest, but look for the last insn that x is modified in
33619 within the same bb.
33621 PR rtl-optimization/80491
33622 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
33624 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
33626 PR tree-optimization/80487
33627 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
33629 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33631 PR tree-optimization/79697
33632 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
33633 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
33634 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
33636 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
33637 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
33639 2017-04-28 Martin Sebor <msebor@redhat.com>
33641 PR tree-optimization/80523
33642 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
33643 (init_target_to_host_charmap, target_to_host, target_strtol10): New
33645 (maybe_warn, format_directive, parse_directive): Use new functions.
33646 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
33648 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
33650 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
33652 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
33654 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
33655 target_header_dir): Set correctly.
33656 * configure: Regenerated.
33657 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33658 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33659 instead of SYSTEM_HEADER_DIR.
33661 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
33663 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
33664 (estimate_local_effects): Likewise.
33665 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
33666 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
33667 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
33668 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
33669 do_estimate_edge_time, estimate_edge_time): Likewise.
33670 * ipa-inline-analysis.c (estimate_node_size_and_time,
33671 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
33672 (estimate_time_after_inlining): Remove.
33674 2017-04-28 Martin Liska <mliska@suse.cz>
33676 * doc/gcov.texi: Enhance documentation of gcov.
33678 2017-04-28 Martin Liska <mliska@suse.cz>
33680 * doc/gcov.texi: Sort options in alphabetic order.
33681 * doc/gcov-dump.texi: Likewise.
33682 * doc/gcov-tool.texi: Likewise.
33683 * gcov.c (print_usage): Likewise.
33684 * gcov-dump.c (print_usage): Likewise.
33685 * gcov-tool.c (print_merge_usage_message): Likewise.
33686 (print_rewrite_usage_message): Likewise.
33687 (print_overlap_usage_message): Likewise.
33689 2017-04-28 Martin Liska <mliska@suse.cz>
33691 PR gcov-profile/53915
33692 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
33694 2017-04-28 Martin Liska <mliska@suse.cz>
33696 PR gcov-profile/79891
33697 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
33698 is marked by compiler as living on a line.
33699 (get_cycles_count): Remove usage of the union.
33700 (output_intermediate_file): Likewise.
33701 (find_source): Fix GNU coding style.
33702 (accumulate_line_counts): Remove old non-all block mode.
33703 (output_lines): Remove usage of the union.
33704 * profile.c (output_location): Include all BBs, even if
33705 belonging to a same line (and file) as a previous BB.
33707 2017-04-28 Martin Liska <mliska@suse.cz>
33709 * gcov.c (process_args): Handle new argument 'w'.
33710 (read_graph_file): Assign ID to BBs.
33711 (output_branch_count): Display BB # if verbose flag is set.
33712 (output_lines): Likewise for arcs.
33713 (print_usage): Add '--verbose' option help.
33714 * doc/gcov.texi: Document --verbose (-w) option.
33716 2017-04-28 Martin Liska <mliska@suse.cz>
33718 * gcov.c (struct block_location_info): New struct.
33719 (process_file): Fill up the new structure.
33720 (read_graph_file): Replace usage of encoding by the newly added
33722 (add_line_counts): Likewise.
33723 (accumulate_line_counts): Remove usage of the union.
33724 (function_info::function_info): New function.
33725 (function_info::~function_info): Likewise.
33726 (process_file): Call delete instead of release_function.
33727 (release_function): Release the function.
33728 (release_structures): Call delete instead of release_function.
33729 (solve_flow_graph): Replace usage of num_blocks.
33730 (find_exception_blocks): Likewise.
33731 (output_lines): Fix GNU coding style.
33733 2017-04-28 Martin Liska <mliska@suse.cz>
33736 * coverage.c (coverage_remove_note_file): New function.
33737 * coverage.h: Declare the function.
33738 * toplev.c (finalize): Clean if an error has been seen.
33740 2017-04-28 Martin Liska <mliska@suse.cz>
33742 PR gcov-profile/80031
33743 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
33744 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
33745 * gcov.c (read_graph_file): Read just number of blocks.
33746 * profile.c (branch_prob): Do not stream 0 flags per a basic
33749 2017-04-28 Martin Liska <mliska@suse.cz>
33751 * gcov-dump.c (tag_*): Add new argument to declarations.
33752 (dump_gcov_file): Likewise.
33753 (tag_blocks): Add and use new argument depth.
33754 (tag_arcs): Likewise.
33755 (tag_lines): Likewise.
33756 (tag_counters): Likewise.
33757 (tag_summary): Likewise.
33758 (dump_working_sets): Use depth to do a proper indentation.
33760 2017-04-28 Jakub Jelinek <jakub@redhat.com>
33763 * cgraph.h (symtab_node::debug_symtab): No longer inline.
33764 * symtab.c (symtab_node::debug_symtab): Move definition here.
33766 2017-04-28 Richard Biener <rguenther@suse.de>
33768 * lto-streamer.h (LTO_major_version): Bump to 7.
33770 2017-04-28 Richard Biener <rguenther@suse.de>
33772 * tree-vrp.c (assert_info): New struct.
33773 (add_assert_info): New helper.
33774 (register_edge_assert_for_2): Refactor to add asserts to a vector
33776 (register_edge_assert_for_1): Likewise.
33777 (register_edge_assert_for): Likewise.
33778 (finish_register_edge_assert_for): New helper actually registering
33779 asserts where live on edge.
33780 (find_conditional_asserts): Adjust.
33781 (find_switch_asserts): Likewise.
33782 (evrp_dom_walker::try_find_new_range): Generalize.
33783 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
33785 2017-04-27 Marek Polacek <polacek@redhat.com>
33788 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
33789 arg10 and arg11 to itype.
33791 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
33793 * doc/extend.texi (Object Size Checking): Improve grammar.
33795 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
33798 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
33799 that the logic for permitting reciprocal estimates matches that
33802 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33805 * tree.c (type_cache_hasher::equal): Only compare
33806 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33807 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33808 non-aggregate element types.
33809 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33810 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33812 2017-04-27 Richard Biener <rguenther@suse.de>
33814 PR middle-end/80533
33815 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33816 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33817 keeping a reference to a trailing array.
33819 2017-04-27 Richard Biener <rguenther@suse.de>
33821 PR middle-end/80539
33822 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33823 being in loop-closed SSA form conservatively.
33824 (chrec_fold_multiply_poly_poly): Likewise.
33826 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33828 PR middle-end/79665
33829 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33830 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33832 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33835 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33836 (aarch64_function_arg_alignment): Return unsigned int again, but still
33837 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33838 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33839 Don't emit -Wpsabi note.
33840 (aarch64_function_arg_boundary): Likewise.
33841 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33844 2017-04-26 Nathan Sidwell <nathan@acm.org>
33846 * tree.h (crc32_unsigned_n): Declare.
33847 (crc32_unsigned, crc32_unsigned): Make inline.
33848 * tree.c (crc32_unsigned_bits): Replace with ...
33849 (crc32_unsigned_n): ... this.
33850 (crc32_unsigned, crc32_byte): Remove.
33851 (crc32_string): Remove unnecessary braces.
33853 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33855 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33856 * ipa-inline-analysis.c (MAX_TIME): Remove.
33857 (account_size_time): Use sreal for time.
33858 (dump_inline_summary): Update.
33859 (estimate_function_body_sizes): Update.
33860 (estimate_edge_size_and_time): Update.
33861 (estimate_calls_size_and_time): Update.
33862 (estimate_node_size_and_time): Update.
33863 (inline_merge_summary): Update.
33864 (inline_update_overall_summary): Update.
33865 (estimate_time_after_inlining): Update.
33866 (inline_read_section): Update.
33867 (inline_write_summary): Update.
33868 * ipa-inline.c (compute_uninlined_call_time): Update.
33869 (compute_inlined_call_time): Update.
33870 (recursive_inlining): Update.
33871 (inline_small_functions): Update.
33872 (dump_overall_stats): Update.
33873 * ipa-inline.h: Include sreal.h.
33874 (size_time_entry): Turn time to sreal.
33875 (inline_summary): Turn self_time nad time to sreal.
33877 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33879 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33881 (sreal::stream_out, sreal::stream_in): New.
33882 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33884 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33886 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33889 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33892 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33894 (dimode_scalar_chain::compute_convert_gain): Ditto.
33895 (dimode_scalar_chain::make_vector_copies): Ditto.
33896 (dimode_scalar_chain::convert_reg): Ditto.
33897 (dimode_scalar_chain::convert_insn): Ditto.
33898 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33899 (VI248_AVX512BW_1): New mode iterator.
33900 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33901 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33904 2017-04-25 Martin Sebor <msebor@redhat.com>
33906 PR tree-optimization/80497
33907 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33908 constants are representable in HOST_WIDE_INT.
33909 (parse_directive): Ditto.
33911 2017-04-25 Martin Sebor <msebor@redhat.com>
33914 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
33915 (new_zero_array): Adjust signature.
33916 (dom_info::dom_init): Used unsigned rather that size_t.
33917 (dom_info::dom_info): Same.
33919 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33920 Jakub Jelinek <jakub@redhat.com>
33923 * config/arm/arm.c: Include gimple.h.
33924 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33925 returns negative, increment ncrn only if it returned positive.
33926 (arm_needs_doubleword_align): Return int instead of bool,
33927 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
33928 members, but if there is any such non-FIELD_DECL
33929 > PARM_BOUNDARY aligned decl, return -1 instead of false.
33930 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33931 returns negative, increment nregs only if it returned positive.
33932 (arm_setup_incoming_varargs): Likewise.
33933 (arm_function_arg_boundary): Emit -Wpsabi note if
33934 arm_needs_doubleword_align returns negative, return
33935 DOUBLEWORD_ALIGNMENT only if it returned positive.
33937 2017-04-25 Marek Polacek <polacek@redhat.com>
33940 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33941 first argument to type.
33943 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
33946 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
33947 type checks to test for compatibility instead of equality.
33949 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33950 Jakub Jelinek <jakub@redhat.com>
33953 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
33955 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
33956 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
33957 the alignment computation, but return their maximum in warn_alignment.
33958 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33959 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
33961 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
33962 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33965 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33967 * config/arc/simdext.md (dmpyh): Fix typo.
33969 2017-04-25 Richard Biener <rguenther@suse.de>
33971 PR tree-optimization/80492
33972 * alias.c (compare_base_decls): Handle registers with asm
33973 specification conservatively.
33974 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33975 compare_base_decls returning dont-know properly.
33977 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33979 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
33980 (legitimate_offset_address_p): New function.
33981 (arc_legitimate_address_p): Use above function.
33983 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33985 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
33987 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33989 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
33990 ACCH registers whenever they are available.
33992 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33994 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
33995 double regs fix when not used.
33997 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33999 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
34001 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
34002 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
34004 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34006 * config/arc/arc.c (arc_output_addsi): Check for h-register class
34007 when emitting short ADD instructions.
34009 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34011 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
34013 (cmpsi_cc_c_insn): Likewise.
34014 (cbranchsi4_scratch): Compute proper instruction length using
34015 compact_hreg_operand.
34016 * config/arc/predicates.md (compact_hreg_operand): New predicate.
34018 2017-04-25 Richard Biener <rguenther@suse.de>
34020 PR middle-end/80509
34021 * passes.c (pass_manager::pass_manager): Initialize
34022 m_name_to_pass_map.
34024 2017-04-25 Richard Biener <rguenther@suse.de>
34026 PR tree-optimization/79201
34027 * tree-ssa-sink.c (statement_sink_location): Handle calls.
34029 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34032 * config/s390/vector.md: Split MEM->GPR vector moves for
34033 non-s_operand addresses.
34035 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34038 * config/s390/predicates.md (reload_const_wide_int_operand): New
34040 * config/s390/s390.md ("movti"): Remove d/P alternative.
34041 ("movti_bigconst"): New pattern definition.
34043 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34046 * s390-protos.h (s390_expand_cs_hqi): Removed.
34047 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
34048 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
34049 modes as well as CCZ1mode and CCZmode.
34050 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
34051 signature of s390_emit_compare_and_swap.
34052 (s390_expand_cs_hqi): Likewise, make static.
34053 (s390_expand_cs_tdsi): Generate an explicit compare before trying
34054 compare-and-swap, in some cases.
34055 (s390_expand_cs): Wrapper function.
34056 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
34058 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
34059 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
34060 patterns for small and large integers. Forbid symref memory operands.
34061 Move expander to s390.c. Require cc register.
34062 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
34063 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
34064 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
34065 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
34066 symref memory operands. Remove CC mode and call s390_match_ccmode
34068 ("atomic_exchange<mode>"): Allow and implement all integer modes.
34070 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34072 * config/s390/s390.md (define_peephole2): New peephole to help
34073 combining the load-and-test pattern with volatile memory.
34075 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34077 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
34078 with CCZmode for TARGET_Z196.
34080 2017-04-25 Jakub Jelinek <jakub@redhat.com>
34082 PR rtl-optimization/80501
34083 * combine.c (make_compound_operation_int): Set subreg_code to SET
34084 even for AND with mask of the sign bit of mode.
34086 PR rtl-optimization/80500
34087 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
34088 sum's initial value.
34090 2017-04-25 Julian Brown <julian@codesourcery.com>
34091 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34093 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
34095 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
34097 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
34099 2017-04-25 Julian Brown <julian@codesourcery.com>
34100 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34102 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
34103 (thunderx2t99_sha): New Reservation.
34105 2017-04-25 Julian Brown <julian@codesourcery.com>
34106 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34108 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
34109 type for 1-element load.
34111 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
34113 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
34115 2017-04-24 Martin Jambor <mjambor@suse.cz>
34117 PR tree-optimization/80293
34118 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
34119 char arrays not totally scalarizable if it is false.
34120 (analyze_all_variable_accesses): Pass correct value in the new
34121 parameter. Add a statistics counter.
34123 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
34125 PR middle-end/79931
34126 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
34128 2017-04-24 Richard Biener <rguenther@suse.de>
34130 PR tree-optimization/80494
34131 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
34132 out for complex types.
34134 2017-04-24 Richard Biener <rguenther@suse.de>
34136 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
34137 * tree-ssa-sccvn.c (print_scc): Print SCC size.
34138 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
34139 (DFS): Adjust and never fail.
34140 (sccvn_dom_walker::fail): Remove.
34141 (sccvn_dom_walker::before_dom_children): Adjust.
34142 (run_scc_vn): Likewise and never fail.
34143 * tree-ssa-pre.c (pass_pre::execute): Adjust.
34144 (pass_fre::execute): Likewise.
34146 2017-04-24 Richard Biener <rguenther@suse.de>
34148 PR tree-optimization/79725
34149 * tree-ssa-sink.c (statement_sink_location): Return whether
34150 failure reason was zero uses. Move that check later.
34151 (sink_code_in_bb): Deal with zero uses by removing the stmt
34154 2017-04-24 Richard Biener <rguenther@suse.de>
34157 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
34158 pointer-based references.
34160 2017-04-24 Richard Biener <rguenther@suse.de>
34163 * pass_manager.h (pass_manager::operator new): Remove.
34164 (pass_manager::operator delete): Likewise.
34165 * passes.c (pass_manager::operator new): Remove.
34166 (pass_manager::operator delete): Likewise.
34167 (pass_manager::pass_manager): Zero individual pass members.
34169 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
34172 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
34173 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
34174 Check "XEXP (src, 1)" operand here.
34175 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
34176 Check "XEXP (src, 1)" operand here.
34177 (dimode_scalar_chain::make_vector_copies): Detect count register
34178 of a shift instruction. Zero extend count register from QImode
34179 to DImode to satisfy vector shift pattern count operand predicate.
34180 Substitute vector shift count operand with a DImode copy.
34181 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
34184 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
34186 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
34187 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34188 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
34189 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34190 (UNSPEC_NOREX_MEM): Remove definition.
34192 2017-04-21 Richard Biener <rguenther@suse.de>
34194 PR tree-optimization/79547
34195 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
34196 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
34197 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
34198 without any constraints.
34200 2017-04-21 Richard Biener <rguenther@suse.de>
34202 PR tree-optimization/78847
34203 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
34205 2017-04-21 Richard Biener <rguenther@suse.de>
34207 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
34208 (build_distinct_type_copy): Likewise.
34209 (build_variant_type_copy): Likewise.
34210 * tree.c (build_qualified_type): Pass down mem-stat info.
34211 (build_distinct_type_copy): Likewise.
34212 (build_variant_type_copy): Likewise.
34214 2017-04-21 Richard Biener <rguenther@suse.de>
34216 PR tree-optimization/80237
34217 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
34219 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
34220 for a simplified result.
34222 2016-04-21 Richard Biener <rguenther@suse.de>
34224 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
34225 sth as strict as a simple_iv but a chrec without symbols and an
34226 operand defined in the loop we are peeling (and not some subloop).
34227 (propagate_constants_for_unrolling): Propagate all constants.
34229 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34232 * config/i386/i386.c (print_reg): Remove assert for disalowed
34233 regno values, call output_operand_lossage instead.
34235 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34238 * config/i386/constraints.md (Yc): New register constraint.
34239 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
34240 Use Yc constraint for alternative 2 of operand 0. Remove
34241 preferred_for_speed attribute.
34243 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34245 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
34246 lastprivate clauses in SIMT case.
34248 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
34250 * doc/invoke.texi (-Wextra-semi): Document new warning option.
34252 2017-04-20 Richard Biener <rguenther@suse.de>
34254 PR tree-optimization/57796
34255 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
34256 as N scalar stores.
34257 (vect_model_load_cost): Cost gathers as N scalar loads.
34259 2017-04-20 Richard Biener <rguenther@suse.de>
34261 * ggc-page.c (ggc_allocated_p): Rename to ...
34262 (safe_lookup_page_table_entry): ... this and return the lookup
34264 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
34266 2017-04-20 Richard Biener <rguenther@suse.de>
34268 PR tree-optimization/80453
34269 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
34270 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
34271 from the conditions.
34272 (vn_phi_eq): Pass them down.
34273 (vn_phi_lookup): Record them.
34274 (vn_phi_insert): Likewise.
34276 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
34278 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
34279 uninitialized variable warning to avoid buffer overrun.
34281 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34284 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
34285 is suppressed for '{ 0 }' in C.
34287 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34289 * BASE-VER: Set to 8.0.0.
34291 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
34293 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
34294 priority .init_array and .fini_array section with SECTION_NOTYPE
34297 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34299 PR middle-end/80423
34300 * tree.h (build_array_type): Add typeless_storage default argument.
34301 * tree.c (type_cache_hasher::equal): Also compare
34302 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
34303 (build_array_type): Add typeless_storage argument, set
34304 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
34306 (build_nonshared_array_type): Adjust build_array_type_1 caller.
34307 (build_array_type): Likewise. Add typeless_storage argument.
34309 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34310 Jakub Jelinek <jakub@redhat.com>
34312 PR tree-optimization/80426
34313 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
34314 operation on symbolic operands, also compute the overflow for the
34315 invariant part when the operation degenerates into a negation.
34317 2017-04-19 Jakub Jelinek <jakub@redhat.com>
34320 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
34321 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
34324 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
34326 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
34329 * config/avr/avr.c (tree.h): Include it.
34330 (cgraph.h): Include it.
34331 (avr_encode_section_info): Don't warn for uninitialized progmem
34332 variable if it's just an alias.
34334 2017-04-19 Richard Biener <rguenther@suse.de>
34337 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
34338 when needed by AutoPGO.
34340 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
34343 * doc/lto.texi: Remove an extra 'that'.
34345 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
34347 PR rtl-optimization/80429
34348 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
34349 are only used in debug insns.
34351 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34352 Vladimir Makarov <vmakarov@redhat.com>
34354 * config/sparc/predicates.md (input_operand): Add comment. Return
34355 true for any memory operand when LRA is in progress.
34356 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
34358 2017-04-18 Jeff Law <law@redhat.com>
34361 * mips.md ({return,simple_return}_internal): Do not overwrite
34364 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34366 PR tree-optimization/80443
34367 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
34368 instead of adding 1, subtract -1 and similarly instead of subtracting
34371 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
34373 PR rtl-optimization/80357
34374 * haifa-sched.c (tmp_bitmap): New variable.
34375 (model_recompute): Handle duplicate use records.
34376 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
34377 (free_global_sched_pressure_data): Free it.
34379 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
34382 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34383 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34384 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34385 instead of SYSTEM_HEADER_DIR.
34387 2017-04-18 Jeff Law <law@redhat.com>
34389 PR middle-end/80422
34390 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
34391 predecessors after walking up the insn chain.
34393 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34396 * dwarf2out.c (modified_type_die): Try harder not to emit internal
34397 sizetype type into debug info.
34399 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34402 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
34403 unneeded test for TARGET_UPPER_REGS_SF.
34404 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
34406 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34409 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
34410 instead of gsi_after_labels.
34412 2017-04-18 Jeff Law <law@redhat.com>
34414 * regcprop.c (maybe_mode_change): Avoid creating copies of the
34418 2017-04-13 Jeff Law <law@redhat.com>
34419 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34420 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34422 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
34425 * config/avr/avr.c (intl.h): Include it.
34426 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
34428 2017-04-18 Martin Liska <mliska@suse.cz>
34430 PR gcov-profile/78783
34431 * gcov-tool.c (gcov_output_files): Validate that destination
34432 file is either removed by the tool or by a user.
34434 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34435 Guy Benyei <guybe@mellanox.com>
34437 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
34438 block, and do not negate it, the stored id is already negative.
34440 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34442 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
34444 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34447 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
34448 masks of options that should be turned off if the VSX vector
34449 options are turned off.
34450 (OTHER_P8_VECTOR_MASKS): Likewise.
34451 (OTHER_VSX_VECTOR_MASKS): Likewise.
34452 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
34453 rs6000_disable_incompatible_switches to validate no type switches
34455 (rs6000_incompatible_switch): New function to disallow turning on
34456 other vector options if -mno-vsx, -mno-power8-vector, or
34457 -mno-power9-vector are specified.
34459 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34461 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
34463 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34465 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
34466 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
34467 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
34468 (ARG_POINTER_CFA_OFFSET): Likewise.
34470 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34472 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
34473 conditions to take advantage of various optimizations.
34475 2017-04-13 Jeff Law <law@redhat.com>
34477 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34478 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34479 (zero_extendsidi2_dext): Likewise.
34481 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34484 * fold-const.c (fold_ternary_loc): Revert
34485 use op0 instead of fold_convert_loc (loc, type, arg0) part of
34488 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
34490 PR rtl-optimization/80343
34491 * lra-remat.c (update_scratch_ops): Assign original hard reg to
34492 new scratch pseudo.
34494 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
34497 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
34498 to ubsan_encode_value.
34500 2017-04-13 Jeff Law <law@redhat.com>
34502 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
34503 appearing in DEBUG_INSNs.
34505 2017-04-13 Martin Liska <mliska@suse.cz>
34507 PR gcov-profile/80413
34508 * gcov-io.c (gcov_write_string): Copy to buffer just when
34509 allocated size is greater than zero.
34511 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34514 * dwarf2out.c (decls_for_scope): Ignore declarations of
34515 current_function_decl in BLOCK_NONLOCALIZED_VARS.
34517 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
34520 * ipa-visibility.c (non_local_p): Fix typos.
34521 (localize_node): When localizing symbol in same comdat group,
34522 dissolve the group only when we know external symbols are going
34524 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
34526 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34528 PR tree-optimization/79390
34529 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
34530 order does not result in usable sequence, retry with reversed operand
34536 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
34537 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
34538 op0 instead of fold_convert_loc (loc, type, arg0).
34540 2017-04-12 Jeff Law <law@redhat.com>
34542 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
34543 has a delay slot in the generated code.
34545 * config/cris/cris.md (cris_preferred_reload_class): Return
34546 GENNONACR_REGS rather than GENERAL_REGS.
34548 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34551 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
34552 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
34553 signedness of the result type.
34555 2017-04-12 Richard Biener <rguenther@suse.de>
34556 Jeff Law <law@redhat.com>
34558 PR tree-optimization/80359
34559 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
34560 trim stores to TARGET_MEM_REFs.
34562 2017-04-12 Richard Biener <rguenther@suse.de>
34564 PR tree-optimization/79390
34565 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
34566 threading case even more.
34568 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
34571 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
34572 for quad_address_p for TImode, instead of just not indexed_address.
34574 2017-04-12 Richard Biener <rguenther@suse.de>
34575 Bernd Edlinger <bernd.edlinger@hotmail.de>
34577 PR middle-end/79671
34578 * alias.c (component_uses_parent_alias_set_from): Handle
34579 TYPE_TYPELESS_STORAGE.
34580 (get_alias_set): Likewise.
34581 * tree-core.h (tree_type_common): Add typeless_storage flag.
34582 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
34583 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
34584 for types containing members with TYPE_TYPELESS_STORAGE.
34585 (place_field): Likewise.
34586 (layout_type): Likewise for ARRAY_TYPE.
34587 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
34588 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
34589 TYPE_TYPELESS_STORAGE.
34590 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
34592 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34595 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
34596 first argument to type.
34598 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34602 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
34603 CONST0_RTX (mode) rather than const0_rtx where appropriate.
34604 (rs6000_expand_binop_builtin): Likewise.
34605 (rs6000_expand_ternop_builtin): Likewise; also add missing
34606 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
34607 vshasigma built-ins.
34608 * doc/extend.texi: Document that vec_xxpermdi's third argument
34609 must be a constant.
34611 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
34613 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
34614 Use shift_const cost parameter when calculating gain of STV shifts.
34616 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
34618 PR rtl-optimization/70478
34619 * lra-constraints.c (process_alt_operands): Check memory for
34620 disfavoring memory insn operand.
34622 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34624 PR middle-end/80100
34625 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
34626 left shift in unsigned HOST_WIDE_INT type.
34628 PR rtl-optimization/80385
34629 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
34630 (not (neg X)) into (plus X -1) for complex or non-integral modes.
34633 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
34634 if they have any depend clauses.
34636 2017-04-11 Martin Liska <mliska@suse.cz>
34639 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
34640 * ipa-split.c (split_function): Create a local comdat symbol
34641 if caller is in a comdat group.
34643 2017-04-11 Martin Liska <mliska@suse.cz>
34646 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
34649 2017-04-11 Martin Sebor <msebor@redhat.com>
34651 PR middle-end/80364
34652 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
34653 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
34655 (directive::set_width, directive::set_precision, format_character):
34657 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
34660 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
34663 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
34664 conflict, set target->arch_name instead of target->cpu_name.
34666 2017-04-11 Richard Biener <rguenther@suse.de>
34668 PR tree-optimization/80374
34669 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
34670 build_zero_cst, remove fold_convertible_p check again.
34672 2017-04-11 Martin Liska <mliska@suse.cz>
34675 * ubsan.c (instrument_object_size): Do not instrument register
34678 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34681 * config/i386/i386-builtin-types.def
34682 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
34683 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
34684 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
34685 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
34686 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
34687 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
34688 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
34689 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
34690 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
34691 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
34692 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
34693 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
34694 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
34695 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
34696 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
34697 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
34698 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
34699 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
34700 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
34701 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
34702 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
34703 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
34704 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
34705 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
34706 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
34707 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
34708 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
34709 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
34710 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
34711 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
34712 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
34713 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
34714 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
34715 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
34716 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
34717 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
34718 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
34719 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
34720 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
34721 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
34722 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
34723 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
34724 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
34725 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
34726 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
34727 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
34729 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
34730 flag to second_arg_count, handle 4 argument function type _COUNT
34731 aliases, handle second_arg_count on second argument rather than last.
34733 2017-04-10 Jeff Law <law@redhat.com>
34735 PR tree-optimization/80374
34736 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
34737 record anything if we can not convert integer_zero_node to the
34740 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34743 * config/rs6000/rs6000.c (rs6000_option_override_internal):
34744 Enhance special handling given to the TARGET_P9_MINMAX option in
34745 relation to certain other options.
34747 2017-04-10 Bin Cheng <bin.cheng@arm.com>
34749 PR tree-optimization/80153
34750 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
34751 remove POINTER_PLUS_EXPR's base part directly, rather than through
34754 2017-04-10 Richard Biener <rguenther@suse.de>
34755 Bin Cheng <bin.cheng@arm.com>
34757 PR tree-optimization/80153
34758 * tree-affine.c (aff_combination_to_tree): Get base pointer from
34759 the first element of pointer type aff_tree. Build result expr in
34761 (add_elt_to_tree): Convert to type unconditionally. Remove other
34762 fold_convert calls.
34763 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
34764 (rewrite_use_nonlinear_expr): Check invariant using iv information.
34766 2017-04-10 Richard Biener <rguenther@suse.de>
34768 * tree-ssa-structalias.c (find_func_aliases): Properly handle
34771 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
34773 PR rtl-optimization/70478
34774 * lra-constraints.c (curr_small_class_check): New.
34775 (update_and_check_small_class_inputs): New.
34776 (process_alt_operands): Update curr_small_class_check. Disfavor
34777 alternative insn memory operands. Check available regs for small
34780 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34783 * config/mips/mips.opt (-mvirt): Update description.
34784 * doc/invoke.texi (-mvirt): Likewise.
34786 2017-04-10 Richard Biener <rguenther@suse.de>
34788 PR middle-end/80362
34789 * fold-const.c (fold_binary_loc): Look at unstripped ops when
34790 looking for NEGATE_EXPR in -A / -B to A / B folding.
34792 2017-04-10 Martin Liska <mliska@suse.cz>
34794 PR gcov-profile/80224
34795 * gcov.c (print_usage): Fix usage string.
34796 (get_gcov_intermediate_filename): Remove.
34797 (output_gcov_file): Use both for normal and intermediate format.
34798 (generate_results): Do not initialize special file for
34799 intermediate format.
34801 2017-04-10 Richard Biener <rguenther@suse.de>
34803 PR tree-optimization/80304
34804 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34807 2017-04-10 Nathan Sidwell <nathan@acm.org>
34810 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34811 (rs6000_init_builtins): Use it.
34813 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34815 * config/arm/arm.md (<mrc>): Add mode to SET source.
34816 (<mrrc>): Likewise.
34818 2017-04-10 Richard Biener <rguenther@suse.de>
34820 PR middle-end/80344
34821 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34823 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34826 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34827 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34828 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34829 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34830 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34831 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34832 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34833 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34834 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34835 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34836 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34837 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34838 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34839 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34840 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34841 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34842 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34843 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
34844 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
34845 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
34846 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34847 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34848 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34850 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34852 PR rtl-optimization/70478
34853 * lra-constraints.c: Reverse the last patch.
34855 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34857 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34858 Add comment for WCHAR_T.
34860 2017-04-08 Martin Liska <mliska@suse.cz>
34863 2017-04-07 Martin Liska <mliska@suse.cz>
34866 * ipa-split.c (split_function): Add function part to a same comdat
34869 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34872 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34874 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34876 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34877 (vec_pairing): ...to this.
34878 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34879 (rs6000_sched_init): Adjust for name change.
34880 (struct rs6000_sched_context): Likewise.
34881 (rs6000_init_sched_context): Likewise.
34882 (rs6000_set_sched_context): Likewise.
34884 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34890 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34892 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34893 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34894 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34896 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34898 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34900 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34902 PR rtl-optimization/70703
34903 * ira-color.c (update_conflict_hard_regno_costs): Use
34904 int64_t instead of HOST_WIDE_INT.
34906 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34908 PR rtl-optimization/70478
34909 * lra-constraints.c (process_alt_operands): Disfavor alternative
34910 insn memory operands.
34912 2017-04-07 Jeff Law <law@redhat.com>
34914 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
34915 CALL and NOTE_INSN_CALL_ARG_LOCATION.
34917 2017-04-07 Martin Liska <mliska@suse.cz>
34920 * config/aarch64/aarch64.c (aarch64_process_target_attr):
34921 Show error message instead of an ICE.
34923 2017-04-07 Martin Liska <mliska@suse.cz>
34926 * ipa-split.c (split_function): Add function part to a same comdat
34929 2017-04-07 Richard Biener <rguenther@suse.de>
34931 PR middle-end/80341
34932 * tree.c (get_unwidened): Also handle ! for_type case for
34934 * convert.c (do_narrow): Split out from ...
34935 (convert_to_integer_1): ... here. Do not pass final truncation
34936 type to get_unwidened for TRUNC_DIV_EXPR.
34938 2017-04-07 Richard Biener <rguenther@suse.de>
34940 * tree-affine.c (wide_int_ext_for_comb): Take type rather
34942 (aff_combination_const): Adjust.
34943 (aff_combination_scale): Likewise.
34944 (aff_combination_add_elt): Likewise.
34945 (aff_combination_add_cst): Likewise.
34946 (aff_combination_convert): Likewise.
34947 (add_elt_to_tree): Likewise. Remove unused argument.
34948 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
34950 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
34952 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
34954 * config/arm/arm.c (arm_default_short_enums): Use
34955 ARM_DEFAULT_SHORT_ENUMS.
34956 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
34958 2017-04-06 Jakub Jelinek <jakub@redhat.com>
34961 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
34962 members with redundant out-of-class redeclaration.
34964 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34967 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
34968 * config/i386/i386.md (*zero_extendsidi2):
34969 Add (?*x,*x) and (?*v,*v) alternatives.
34971 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34974 * config/i386/i386.c (ix86_expand_builtin)
34975 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
34976 mode from insn data. Convert operands to insn operand mode.
34977 Copy operands that don't satisfy insn predicate to a register.
34979 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
34981 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
34984 2017-04-06 Richard Biener <rguenther@suse.de>
34986 PR tree-optimization/80334
34987 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
34988 preserve alignment of accesses.
34990 2017-04-06 Richard Biener <rguenther@suse.de>
34992 PR tree-optimization/80262
34993 * tree-sra.c (build_ref_for_offset): Preserve address-space
34995 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
34996 Drop useless address-space information on MEM_REF offsets.
34998 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
35000 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
35002 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
35004 PR rtl-optimization/70703
35005 * ira-color.c (update_conflict_hard_regno_costs): Use
35006 HOST_WIDE_INT instead of long.
35008 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
35011 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
35012 not defined for x86_64 target. Add -mmmx target option when __SSE2__
35014 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
35015 for x86_64 target. Handle -m3dnowa option.
35017 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
35019 PR rtl-optimization/70703
35020 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
35021 (update_conflict_hard_regno_costs): Use long instead of unsigned
35022 arithmetic for cost calculation.
35024 2017-04-05 Jakub Jelinek <jakub@redhat.com>
35025 Bernd Edlinger <bernd.edlinger@hotmail.de>
35028 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
35031 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
35034 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
35035 ptr_mode with Pmode throughout.
35036 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
35037 into probe_stack_range and use DImode.
35039 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
35042 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
35043 call_eh_return is true.
35045 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35047 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
35048 Initialize last_match_fntype_index.
35050 2017-04-05 Jakub Jelinek <jakub@redhat.com>
35053 * tree-nvr.c: Include internal-fn.h.
35054 (pass_return_slot::execute): Ignore internal calls without
35057 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35058 Richard Biener <rguenther@suse.de>
35061 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
35062 captures used multiple times, except for the last use.
35063 * generic-match-head.c: Include gimplify.h.
35065 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35067 PR tree-optimization/79390
35068 * target.h (struct noce_if_info): Declare.
35069 * targhooks.h (default_noce_conversion_profitable_p): Declare.
35070 * target.def (noce_conversion_profitable_p): New target hook.
35071 * ifcvt.h (struct noce_if_info): New type, moved from ...
35072 * ifcvt.c (struct noce_if_info): ... here.
35073 (noce_conversion_profitable_p): Renamed to ...
35074 (default_noce_conversion_profitable_p): ... this. No longer
35076 (noce_try_store_flag_constants, noce_try_addcc,
35077 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
35078 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
35079 instead of noce_conversion_profitable_p.
35080 * config/i386/i386.c: Include ifcvt.h.
35081 (ix86_option_override_internal): Don't override
35082 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
35083 (ix86_noce_conversion_profitable_p): New function.
35084 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
35085 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
35086 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
35087 * doc/tm.texi: Regenerated.
35089 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35091 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
35094 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
35097 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
35098 instructions for small multiply cores.
35100 2017-04-04 Jeff Law <law@redhat.com>
35102 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
35104 (mips_expand_vec_perm_const): Initialize elements in orig_perm
35105 that are not set by the loop over the elements.
35107 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35110 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
35111 int mode, convert_modes it to mode as unsigned, otherwise use
35112 lowpart_subreg to mode rather than SImode.
35113 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
35114 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
35115 Use DImode instead of SImode for the shift count operand.
35116 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
35119 2017-04-04 Richard Biener <rguenther@suse.de>
35121 PR middle-end/80281
35122 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
35123 arithmetic done for the negate or the plus. Simplify.
35124 (A - (-B) -> A + B): Likewise.
35125 * fold-const.c (split_tree): Make sure to not negate pointers.
35127 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
35129 PR rtl-optimization/60818
35130 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
35131 a compare of comparisons with the thing compared if this results
35132 in a different machine mode.
35134 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
35136 * alias.c (base_alias_check): Fix typo in comment.
35137 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
35138 * cgraphunit.c (symbol_table::compile): Likewise.
35139 * collect2.c (maybe_run_lto_and_relink): Likewise.
35140 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
35141 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
35142 * config/avr/avr.c (avr_map_op_t): Likewise.
35143 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
35144 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
35145 * config/epiphany/epiphany.md (movcc): Likewise.
35146 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
35147 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
35149 * config/mips/mips.c (mips_save_restore_reg): Likewise.
35150 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
35151 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
35152 * config/sh/sh.c (sh_rtx_costs): Likewise.
35153 * fold-const.c (fold_truth_andor): Likewise.
35154 * genautomata.c (collapse_flag): Likewise.
35155 * gengtype.h (struct type::u::s): Likewise.
35156 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
35157 * input.c (FORMAT_AMOUNT): Likewise.
35158 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
35159 (known_aggs_to_agg_replacement_list): Likewise.
35160 * ipa-inline-analysis.c: Likewise.
35161 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
35162 * ipa-polymorphic-call.c
35163 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
35164 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
35165 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
35167 * modulo-sched.c (apply_reg_moves): Likewise.
35168 * omp-expand.c (build_omp_regions_1): Likewise.
35169 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
35170 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
35171 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
35172 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
35173 * value-prof.c: Likewise.
35174 * var-tracking.c (val_reset): Likewise.
35176 2017-04-03 Richard Biener <rguenther@suse.de>
35178 PR tree-optimization/80275
35179 * fold-const.c (split_address_to_core_and_offset): Handle
35182 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
35184 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
35185 descriptors is at least equal to that of functions.
35187 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35189 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
35191 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35194 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
35195 (mov<IMOD4:mode>): New expander.
35196 (*mov<IMOD4:mode>_internal): New insn and split pattern.
35198 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
35200 PR rtl-optimization/79405
35201 * fwprop.c (propagations_left): New variable.
35202 (forward_propagate_into): Decrement it.
35203 (fwprop_init): Initialize it.
35204 (fw_prop): If the variable has reached zero, stop propagating.
35205 (fwprop_addr): Ditto.
35207 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35210 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
35211 a FUNCTION_DECL, pass it as decl instead of origin to
35214 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
35216 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
35219 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
35222 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
35223 TARGET_VSX_SMALL_INTEGER.
35225 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35227 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
35228 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
35230 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
35232 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
35233 extraction from odd-numbered MSA register.
35235 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35237 PR middle-end/80173
35238 * expmed.c (store_bit_field_1): Don't attempt to create
35239 a word subreg out of hard registers wider than word if they
35240 have HARD_REGNO_NREGS of 1 for their mode.
35242 PR middle-end/80163
35243 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
35244 conversions to integer types wider than word and pointer.
35247 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
35248 (rtx_equal_for_cselib_p): Pass 0 to it.
35249 * cselib.c (cselib_hasher::equal): Likewise.
35250 (rtx_equal_for_cselib_1): Add depth argument. If depth
35251 is 128, don't look up VALUE locs and punt. Increment
35252 depth in recursive calls when walking VALUE locs.
35254 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
35256 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
35257 (make_gcov_file_name): Use the canonical path name for generating
35259 (read_line): Fix handling of files with ascii null bytes.
35261 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
35263 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
35264 to initialise a vector register instead
35265 of using a const_int.
35267 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35269 PR translation/80189
35270 * gimplify.c (omp_default_clause): Use %qs instead of %s in
35271 diagnostic messages.
35273 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
35276 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
35277 (dfp_diex_<mode>): Update mode of operand 1.
35278 * doc/extend.texi (dxex, dxexq): Document change to return type.
35279 (diex, diexq): Document change to argument type.
35281 2017-03-30 Martin Jambor <mjambor@suse.cz>
35284 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
35285 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
35286 it reflects the signature changes performed at the callee side.
35287 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
35288 to cgraph_build_function_type_skip_args.
35289 (build_function_decl_skip_args): Adjust call to the above function.
35291 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35294 * config/i386/sse.md
35295 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
35296 register as dest whenever it is a MEM not rtx_equal_p to the
35297 corresponding dup operand, and when forcing into reg move the
35298 reg into the memory afterwards.
35299 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
35300 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
35301 for the force_reg mode.
35302 (avx512vl_vextractf128<mode>): Use register as dest either
35303 always when a MEM, or when it is a MEM not rtx_equal_p to the
35304 corresponding dup operand, or even not when it is a CONST_VECTOR
35305 depending on the mode and lo vs. hi.
35306 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
35308 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
35309 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
35310 Likewise. Require that operands[2] is even.
35311 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
35312 Remove extraneous parens. Require that operands[2] is a multiple
35314 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
35315 operands[0] is a MEM if <mask_applied>, the predicates/constraints
35316 disallow memory then.
35318 2017-03-30 Richard Biener <rguenther@suse.de>
35320 PR tree-optimization/77498
35321 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
35322 to non-constants over backedges.
35324 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
35326 PR rtl-optimization/80233
35327 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
35328 as last_combined_insn. Do not test for BARRIER_P separately.
35330 2017-03-29 Andreas Schwab <schwab@suse.de>
35333 * calls.c (prepare_call_address): Convert funexp to Pmode before
35334 copying to temp reg.
35336 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35338 PR tree-optimization/80158
35339 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
35340 Handle possible future case of more than one alternate
35342 (replace_rhs_if_not_dup): Likewise.
35343 (replace_one_candidate): Likewise.
35345 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
35347 PR rtl-optimization/80193
35348 * ira.c (ira): Do not check allocation for LRA.
35350 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
35352 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
35353 (nvptx_output_simt_exit): Declare.
35354 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
35355 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
35356 (init_softstack_frame): Move initialization of crtl->is_leaf to...
35357 (nvptx_declare_function_name): ...here. Emit declaration of local
35358 memory space buffer for omp_simt_enter insn.
35359 (nvptx_output_unisimt_switch): New.
35360 (nvptx_output_softstack_switch): New.
35361 (nvptx_output_simt_enter): New.
35362 (nvptx_output_simt_exit): New.
35363 * config/nvptx/nvptx.h (struct machine_function): New fields
35364 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
35365 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
35366 (UNSPECV_SIMT_EXIT): Ditto.
35367 (omp_simt_enter_insn): New insn.
35368 (omp_simt_enter): New expansion.
35369 (omp_simt_exit): New insn.
35370 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
35372 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
35373 (expand_GOMP_SIMT_ENTER_ALLOC): New.
35374 (expand_GOMP_SIMT_EXIT): New.
35375 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
35376 (GOMP_SIMT_ENTER_ALLOC): Ditto.
35377 (GOMP_SIMT_EXIT): Ditto.
35378 * target-insns.def (omp_simt_enter): New insn.
35379 (omp_simt_exit): Ditto.
35380 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
35382 (lower_rec_simd_input_clauses): Implement SIMT privatization.
35383 (lower_rec_input_clauses): Likewise.
35384 (lower_lastprivate_clauses): Handle SIMT privatization.
35386 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
35387 (ompdevlow_adjust_simt_enter): New.
35388 (find_simtpriv_var_op): New.
35389 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
35390 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
35392 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
35393 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
35394 (copy_decl_for_dup_finish): Ditto.
35396 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
35398 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
35401 * config/i386/i386.c (ix86_option_override_internal): Always
35402 allow -mpreferred-stack-boundary=3 for 64-bit targets.
35404 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35406 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
35408 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35410 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
35411 mark new edge's irreducible flag accordign to it.
35412 (vect_do_peeling): Check loop preheader edge's irreducible flag
35413 and pass it to function slpeel_add_loop_guard.
35415 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
35417 PR tree-optimization/80218
35418 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
35419 Update block frequencies and counts.
35421 2017-03-28 Richard Biener <rguenther@suse.de>
35423 PR tree-optimization/78644
35424 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
35425 of a simplification result we may not use it at all.
35427 2017-03-28 Richard Biener <rguenther@suse.de>
35430 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
35431 without arguments, generate default definition of a SSA name.
35433 2017-03-28 Richard Biener <rguenther@suse.de>
35435 PR middle-end/80222
35436 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
35437 TYPE_REF_CAN_ALIAS_ALL references.
35438 * fold-const.c (fold_indirect_ref_1): Likewise.
35440 2017-03-28 Martin Liska <mliska@suse.cz>
35443 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
35444 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
35446 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35447 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
35449 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
35450 (EXTRA_SPECS): Define.
35451 (SUBTARGET_EXTRA_SPECS): Likewise.
35452 (SUBTARGET_CPP_SPEC): Likewise.
35453 * config/arc/elf.h (EXTRA_SPECS): Renamed to
35454 SUBTARGET_EXTRA_SPECS.
35455 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
35457 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35459 * config/arc/simdext.md (vst64_insn): Update pattern.
35460 (vld32wh_insn): Likewise.
35461 (vld32wl_insn): Likewise.
35462 (vld64_insn): Likewise.
35463 (vld32_insn): Likewise.
35465 2017-03-28 Marek Polacek <polacek@redhat.com>
35468 * fold-const.c (fold_comparison): Use protected_set_expr_location
35469 instead of SET_EXPR_LOCATION.
35471 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
35473 * tree.c (add_expr): Avoid name lookup warning.
35475 2017-03-27 Jeff Law <law@redhat.com>
35477 PR tree-optimization/80216
35478 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
35479 function name. Limit recursion depth.
35480 (record_temporary_equivalences): Corresponding changes.
35482 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
35484 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
35487 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35490 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
35492 * cfgcleanup.c (reg_note_cfa_p): New array.
35493 (insns_have_identical_cfa_notes): New function.
35494 (old_insns_match_p): Don't cross-jump in between /f
35495 and non-/f instructions. If both i1 and i2 are frame related,
35496 verify all CFA notes, their order and content.
35498 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
35501 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
35502 HImode and SImode with zero extend to DImode to one insn.
35503 (bswap<mode>2_extenddi): Likewise.
35504 (bswapsi2_extenddi): Likewise.
35505 (bswaphi2_extendsi): Likewise.
35506 (bswaphi2): Combine bswap HImode and SImode into one insn.
35507 Separate memory insns from swapping register.
35508 (bswapsi2): Likewise.
35509 (bswap<mode>2): Likewise.
35510 (bswaphi2_internal): Delete, no longer used.
35511 (bswapsi2_internal): Likewise.
35512 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
35513 store, and gpr<-gpr swap insns.
35514 (bswap<mode>2_store): Likewise.
35515 (bswaphi2_reg): Register only splitter, combine with the splitter.
35516 (bswaphi2 splitter): Likewise.
35517 (bswapsi2_reg): Likewise.
35518 (bswapsi2 splitter): Likewise.
35519 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
35520 the insns into load, store, and register/register insns.
35521 (bswapdi2_ldbrx): Likewise.
35522 (bswapdi2_load): Likewise.
35523 (bswapdi2_store): Likewise.
35524 (bswapdi2_reg): Likewise.
35526 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
35528 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
35529 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
35531 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
35534 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
35536 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35537 special handling for target option conflicts between dform
35538 options (-mpower9-dform, -mpower9-dform-vector,
35539 -mpower9-dform-scalar) and -mno-direct-move.
35541 2017-03-27 Richard Biener <rguenther@suse.de>
35543 PR tree-optimization/80181
35544 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
35546 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35548 * config/arc/predicates.md (move_double_src_operand): Replace the
35549 call to move_double_src_operand with a call to address_operand.
35551 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35553 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
35554 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
35555 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
35557 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35559 * config/arc/predicates.md (long_immediate_loadstore_operand):
35560 Consider scaled addresses cases.
35562 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35564 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
35565 restored when in interrupt.
35566 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
35567 doesn't have delay slot.
35569 2017-03-27 Richard Biener <rguenther@suse.de>
35572 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
35573 inlined thunk clones.
35575 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35578 * asan.c (instrument_derefs): Copy over last operand from
35579 original COMPONENT_REF to the new COMPONENT_REF with
35580 DECL_BIT_FIELD_REPRESENTATIVE.
35581 * ubsan.c (instrument_object_size): Likewise.
35583 2017-03-27 Richard Biener <rguenther@suse.de>
35585 PR tree-optimization/80170
35586 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
35587 sure DR/SCEV didnt fold in constants we do not see when looking
35588 at the reference base alignment.
35590 2017-03-27 Richard Biener <rguenther@suse.de>
35592 PR middle-end/80171
35593 * gimple-fold.c (fold_ctor_reference): Properly guard against
35594 NULL return value from canonicalize_constructor_val.
35596 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
35599 * config/i386/i386.c (ix86_expand_builtin)
35600 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
35601 flags reg setting and flags reg using instructions.
35602 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
35603 clobbering instructions to zero extend op2.
35605 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
35607 * doc/install.texi (Configuration) <--with-aix-soname>:
35608 Update link to AIX ld.
35610 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
35612 PR rtl-optimization/80160
35613 PR rtl-optimization/80159
35614 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
35615 reg_alternate_class into account.
35617 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
35620 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
35621 to consider in curr_insn_transform.
35623 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35625 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
35626 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
35627 and emit_mode_inner.
35629 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35631 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
35632 argument to the overloaded builtin variants. Use the new flag to
35633 deprecate certain builtin variants.
35634 * config/s390/s390-builtin-types.def: Add new builtin types.
35635 * config/s390/s390-builtins.h: Support new flags field for
35636 overloaded builtins.
35637 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
35638 (s390_macro_to_expand): Enable vector float data type.
35639 (s390_cpu_cpp_builtins_internal): Indicate support of the new
35640 builtins by incrementing the __VEC__ version number.
35641 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
35643 (s390_resolve_overloaded_builtin): Emit error messages depending
35644 on the builtin flags.
35645 * config/s390/s390.c (s390_expand_builtin): Support additional
35646 flags argument. Change error message to match the messages
35647 emitted in s390-c.c.
35648 * config/s390/s390.md: New UNSPEC_* constants.
35649 (op_type): Add new instruction types.
35650 * config/s390/vecintrin.h: Add new builtins and test data class
35652 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
35653 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
35654 (VEC_INEXACT, VEC_NOINEXACT): New constants.
35655 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
35656 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
35657 ("vec_mergel<mode>"): V_HW -> VEC_HW.
35659 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
35660 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
35661 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
35662 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
35664 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
35665 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
35666 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
35667 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
35669 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
35670 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
35671 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
35672 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
35673 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
35674 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
35675 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
35677 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
35678 ("vec_scatter_element<V_HW_4:mode>_DI")
35679 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
35680 ("vec_fpint<mode>", "vflls")
35681 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
35682 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
35683 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
35684 ("*vec_cmphe<mode>_cc"): ... these.
35686 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
35687 mode constant instead of magic value.
35689 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35691 * config/s390/s390.c (s390_expand_vec_compare): Support other
35692 vector floating point modes than just V2DF.
35693 (s390_expand_vcond): Likewise.
35694 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
35695 (s390_cannot_change_mode_class): Prevent mode changes between TF
35696 and V1TF in vector registers.
35697 * config/s390/s390.md (DF, SF): New mode attributes.
35698 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
35699 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
35700 SFmode support for VRs.
35701 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
35703 (VFT, VF_HW): New mode iterators.
35704 (vw, sdx): New mode attributes.
35705 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
35706 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
35707 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
35708 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
35709 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
35710 also the new vector floating point modes. Renaming to ...
35712 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
35713 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
35714 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
35715 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
35716 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
35717 ("vec_unordered<mode>"): ... these.
35719 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
35720 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
35721 ("*vec_extendv2df"): New insn definitions.
35723 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35725 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
35726 ("mulditi3_2", "*muldi3_sign"): New patterns.
35727 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
35728 rename the pattern definition.
35730 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35732 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
35734 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
35736 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35738 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
35739 instruction if possible.
35740 * config/s390/vector.md (vec_halfnumelts): New mode
35742 ("*vec_vllezlf<mode>"): New pattern.
35744 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35746 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
35747 ("popcountv4si2", "popcountv2di2"): Rename to ...
35748 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
35749 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
35751 ("popcount<mode>2_vxe"): New pattern.
35753 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35755 * common/config/s390/s390-common.c (processor_flags_table): Add
35757 * config.gcc: Add arch12.
35758 * config/s390/driver-native.c (s390_host_detect_local_cpu):
35759 Default to arch12 for unknown CPU model numbers.
35760 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
35761 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
35762 PROCESSOR_max sanity check.
35763 * config/s390/s390-opts.h (enum processor_type): Add
35765 * config/s390/s390.c (processor_table): Add arch12.
35766 (s390_expand_builtin): Add check for B_VXE flag.
35767 (s390_issue_rate): Add PROCESSOR_ARCH12.
35768 (s390_get_sched_attrmask): Likewise.
35769 (s390_get_unit_mask): Likewise.
35770 (s390_sched_score): Enable z13 scheduling for arch12.
35771 (s390_sched_reorder): Likewise.
35772 (s390_sched_variable_issue): Likewise.
35773 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
35775 (s390_tune_attr): Use z13 scheduling also for arch12.
35776 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
35777 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
35778 (TARGET_VXE_P): New macros.
35779 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
35780 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
35781 * config/s390/s390.opt: Add arch12 as processor_type.
35783 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35785 * config/s390/s390.md
35786 ("fixuns_truncdddi2", "fixuns_trunctddi2")
35787 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
35788 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
35790 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
35791 Rename expanders to ...
35793 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
35794 ("fixuns_truncdddi2_emu"): ... these.
35796 ("fixuns_trunc<mode>si2_emu"): New expander.
35798 ("*fixuns_truncdfdi2_z13"): Rename to ...
35799 ("*fixuns_truncdfdi2_vx"): ... this.
35801 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35803 * config/s390/2964.md: Remove the single element vector compare
35804 instructions which are no longer used.
35805 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35807 (s390_canonicalize_comparison): Remove handling of DFmode
35809 (s390_expand_vec_compare_scalar): Remove function.
35810 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35811 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35813 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35815 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35817 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35819 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35820 will anyway by matched by mov<mode>_64dfp.
35822 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35824 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35825 vlef/vstef. Add missing operand to vleif.
35827 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35829 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35830 pair for all vector types with 64 bit elements.
35831 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35832 * config/s390/vector.md (V_HW_64): ... here.
35833 (V_128_NOSINGLE): New mode iterator.
35834 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35835 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35836 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35837 ("*vec_load_pairv2di"): Change to ...
35838 ("*vec_load_pair<mode>"): ... this one.
35840 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35842 * config/s390/constraints.md: Add comments.
35843 (jKK): Reject element sizes > 8 bytes.
35844 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
35846 * config/s390/s390.md: Add the s_operand checks formerly in
35847 s390_split_ok_p to various splitters where they are still
35849 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35850 for 128 bit vectors. Plus two splitters.
35852 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35854 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35857 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35860 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35861 error if the boundary argument is not constant.
35863 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35865 PR rtl-optimization/80112
35866 * loop-doloop.c (doloop_condition_get): Don't check condition
35867 if cmp isn't SET with IF_THEN_ELSE src.
35869 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35871 PR tree-optimization/80158
35872 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35873 replacing a candidate statement, also replace it for the
35874 candidate's alternate interpretation.
35875 (replace_rhs_if_not_dup): Likewise.
35876 (replace_one_candidate): Likewise.
35878 2017-03-24 Richard Biener <rguenther@suse.de>
35880 PR tree-optimization/80167
35881 * graphite-isl-ast-to-gimple.c
35882 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35884 (translate_isl_ast_to_gimple::get_rename): Likewise.
35886 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35888 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35889 handling of certain combinations of target options, including the
35890 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35891 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35893 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35896 * config/arm/arm.md (*load_multiple): Add reload_completed to
35897 matching condition.
35899 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35900 Richard Biener <rguenth@suse.de>
35902 PR tree-optimization/79908
35903 PR tree-optimization/80136
35904 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35905 been cast away, gimplify_and_add suffices.
35907 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35909 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35911 2017-03-23 Richard Biener <rguenther@suse.de>
35913 PR tree-optimization/80032
35914 * gimplify.c (gimple_push_cleanup): Forced unconditional
35915 cleanups still have to go to the conditional_cleanups
35918 2017-03-22 Jakub Jelinek <jakub@redhat.com>
35920 PR tree-optimization/80072
35921 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
35923 (next_operand_entry_id): Change type to unsigned int.
35924 (sort_by_operand_rank): Make sure to return the right return value
35925 even if unsigned fields are bigger than INT_MAX.
35926 (struct oecount): Change cnt and id type to unsigned int.
35927 (oecount_hasher::equal): Formatting fix.
35928 (oecount_cmp): Make sure to return the right return value
35929 even if unsigned fields are bigger than INT_MAX.
35930 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
35933 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
35934 TREE_READONLY on result if writing it more than once.
35937 * doc/invoke.texi (-fsanitize=thread): Document that with
35938 -fnon-call-exceptions atomics are not able to throw
35942 * tsan.c: Include tree-eh.h.
35943 (instrument_builtin_call): Call maybe_clean_eh_stmt or
35944 maybe_clean_or_replace_eh_stmt where needed.
35945 (instrument_memory_accesses): Add cfg_changed argument.
35946 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
35947 if it returned true.
35948 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
35950 PR rtl-optimization/63191
35951 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
35952 wrapper function, moved the whole old content into ...
35953 (ix86_delegitimize_address_1): ... this. New inline function.
35954 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
35955 true as last argument instead of ix86_delegitimize_address.
35957 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35959 * config/aarch64/aarch64.c (generic_branch_cost): Copy
35960 cortexa57_branch_cost.
35962 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35964 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
35966 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35969 * doc/md.texi (Constraints): Document wA constraint.
35970 * config/rs6000/constraints.md (wA): New.
35971 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
35972 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
35973 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
35974 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
35976 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
35979 * gimplify.c (is_oacc_declared): New function.
35980 (oacc_default_clause): Use it to set default flags for acc declared
35981 variables inside parallel regions.
35982 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
35983 declared variables.
35984 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
35985 declare attribute to any decl as necessary.
35987 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
35990 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
35991 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
35992 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
35993 (arm_arch_lpae): This.
35994 * config/arm/arm.c (arm_arch7ve): Rename into ...
35995 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
35996 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
35999 2017-03-22 Martin Liska <mliska@suse.cz>
36002 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
36003 error message instead of an ICE.
36005 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36007 * doc/extend.texi (6.11 Additional Floating Types): Revise.
36009 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
36011 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36013 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
36016 2017-03-21 Martin Sebor <msebor@redhat.com>
36018 * doc/extend.texi: Use "cannot" instead of "can't."
36019 * doc/hostconfig.texi: Same.
36020 * doc/install.texi: Same.
36021 * doc/invoke.texi: Same.
36022 * doc/loop.texi: Same.
36023 * doc/md.texi: Same.
36024 * doc/objc.texi: Same.
36025 * doc/rtl.texi: Same.
36026 * doc/tm.texi: Same.
36027 * doc/tm.texi.in: Same.
36028 * doc/trouble.texi: Same.
36030 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
36033 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
36034 (collect_checksum_attributes): Set it.
36035 (die_checksum_ordered): Use it.
36037 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36039 PR tree-optimization/79908
36040 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
36041 change: For a VA_ARG whose LHS has been cast away, use
36042 force_gimple_operand to construct the side effects.
36044 2017-03-21 David Malcolm <dmalcolm@redhat.com>
36046 PR translation/80001
36047 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
36048 more amenable to translation.
36049 (oacc_loop_auto_partitions): Likewise.
36051 2017-03-21 Marek Polacek <polacek@redhat.com>
36052 Martin Sebor <msebor@redhat.com>
36054 PR tree-optimization/80109
36055 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
36056 on INTEGRAL_TYPE_P.
36058 2017-03-21 Jakub Jelinek <jakub@redhat.com>
36059 Segher Boessenkool <segher@kernel.crashing.org>
36062 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
36063 check reg_used_between_p between insn and one of succ or succ2
36064 depending on if succ is artificial insn not inserted into insn
36067 2017-03-21 Martin Liska <mliska@suse.cz>
36069 PR gcov-profile/80081
36070 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
36071 * doc/gcc.texi: Include gcov-dump stuff.
36072 * doc/gcov-dump.texi: New file.
36074 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
36076 PR rtl-optimization/79150
36077 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
36078 conditional jump, if the jump is the last insn of the loop.
36080 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36081 Richard Biener <rguenth@suse.de>
36083 PR tree-optimization/79908
36084 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
36085 been cast away, use force_gimple_operand to construct the side
36088 2017-03-21 Martin Liska <mliska@suse.cz>
36090 PR libfortran/79956
36091 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
36094 2017-03-21 Brad Spengler <spender@grsecurity.net>
36097 * plugin.c (htab_hash_plugin): New function.
36098 (add_new_plugin): Use it and adjust.
36099 (parse_plugin_arg_opt): Adjust.
36100 (init_one_plugin): Likewise.
36102 2017-03-21 Richard Biener <rguenther@suse.de>
36104 PR tree-optimization/80032
36105 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
36106 if set force the cleanup to happen unconditionally.
36107 (gimplify_target_expr): Push inserted clobbers with force_uncond
36108 to avoid them being removed by control-dependent DCE.
36110 2017-03-21 Richard Biener <rguenther@suse.de>
36112 PR tree-optimization/80122
36113 * tree-inline.c (copy_bb): Do not expans va-arg packs or
36114 va_arg_pack_len when the inlined call stmt requires pack
36116 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
36118 2017-03-21 Jakub Jelinek <jakub@redhat.com>
36121 * tsan.c (instrument_builtin_call): If the memory model argument
36122 is not a constant, assume it is valid.
36125 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
36128 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36130 PR rtl-optimization/79910
36131 * combine.c (can_combine_p): Do not allow combining an I0 or I1
36132 if its dest is used by an insn before I2 (other than the combined
36133 insns themselves, which are properly handled already).
36135 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36138 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36140 * combine.c (record_used_regs): New static function.
36141 (try_combine): Handle situations where there is an additional
36142 instruction between I2 and I3 which needs to have a LOG_LINK
36146 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36148 * combine.c (try_combine): Delete redundant i1 test. Call
36149 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36151 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36154 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
36155 alternatives 13/14.
36157 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36159 PR tree-optimization/80054
36160 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
36161 the optimization if a PHI or any of its arguments is not dominated
36162 by the candidate's basis. Use gphi* rather than gimple* as
36164 (replace_profitable_candidates): Clean up a gimple* variable that
36165 should be a gphi* variable.
36167 2017-03-20 Martin Sebor <msebor@redhat.com>
36170 * doc/extend.texi (attribute constructor): Document present limitation.
36172 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
36175 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
36176 __POWER9_VECTOR__ #ifdef control, change template definition to
36177 use Power9-specific built-in function.
36178 (vec_any_eq): Likewise.
36179 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
36180 to control outcomes from this test.
36181 (vector_ae_<mode>p): For VEC_F modes, likewise.
36183 2017-03-20 Ian Lance Taylor <iant@google.com>
36185 * config/i386/i386.c (ix86_function_regparm): Save an extra
36186 register for -fsplit-stack with DECL_STATIC_CHAIN.
36188 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36191 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
36192 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
36194 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36196 * config/riscv/riscv.c (riscv_print_operand): Use "fence
36198 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
36201 2017-03-20 Marek Polacek <polacek@redhat.com>
36204 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
36206 2017-03-20 Richard Biener <rguenther@suse.de>
36208 PR tree-optimization/80113
36209 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
36210 allocate extra SSA name for PHI def.
36211 (add_close_phis_to_outer_loops): Likewise.
36212 (add_close_phis_to_merge_points): Likewise.
36213 (copy_loop_close_phi_args): Likewise.
36214 (copy_cond_phi_nodes): Likewise.
36216 2017-03-20 Martin Liska <mliska@suse.cz>
36218 PR middle-end/79753
36219 * tree-chkp.c (chkp_build_returned_bound): Do not build
36220 returned bounds for a LHS that's not a BOUNDED_P type.
36222 2017-03-20 Martin Liska <mliska@suse.cz>
36226 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
36227 COMPLEX_CST and VECTOR_CST.
36229 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36232 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
36233 target operand. A new splitter adds the clobber statement in case
36234 the target operand is dead anyway.
36236 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
36238 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
36239 to age-old versions of binutils and glibc.
36241 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
36243 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
36245 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36247 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
36249 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36251 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
36252 requirement for binutils 2.13.
36254 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36256 * combine.c (try_combine): Delete redundant i1 test. Call
36257 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36259 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
36261 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
36262 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
36264 <riscv64-*-elf>: Re-arrange section
36265 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
36266 <riscv32-*-linux>: Likewise.
36267 <riscv64-*-elf>: Likewise
36268 <riscv64-*-linux>: Likewise.
36270 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
36273 * aarch64.opt(verbose-cost-dump): Fix typo.
36275 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
36278 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
36279 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
36281 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36283 * reload.c (find_reloads): When reloading a nonoffsettable address,
36284 use RELOAD_OTHER for it and its address reloads.
36286 PR rtl-optimization/79910
36287 * combine.c (record_used_regs): New static function.
36288 (try_combine): Handle situations where there is an additional
36289 instruction between I2 and I3 which needs to have a LOG_LINK
36292 2017-03-17 Jeff Law <law@redhat.com>
36294 PR tree-optimization/71437
36295 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
36296 conditional in the hash table first.
36297 (vrp_dom_walker::before_dom_children): Extract condition from
36298 ASSERT_EXPR. Record condition, its inverion and any implied
36299 conditions as well.
36301 2017-03-17 Marek Polacek <polacek@redhat.com>
36302 Markus Trippelsdorf <markus@trippelsdorf.de>
36304 PR tree-optimization/80079
36305 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
36308 2017-03-17 Richard Biener <rguenther@suse.de>
36310 PR middle-end/80075
36311 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
36312 Properly verify the LHS before the RHS possibly claims to be
36314 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
36317 2017-03-17 Martin Jambor <mjambor@suse.cz>
36319 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
36320 (List of -O2 options): Likewise.
36321 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
36324 2017-03-17 Tom de Vries <tom@codesourcery.com>
36326 * gcov-dump.c (print_usage): Print bug_report_url.
36328 2017-03-17 Richard Biener <rguenther@suse.de>
36330 PR middle-end/80050
36331 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
36332 (parser::peek): Likewise.
36334 2017-03-17 Richard Biener <rguenther@suse.de>
36336 PR tree-optimization/80048
36337 * sese.c (free_sese_info): Properly release rename_map and
36338 copied_bb_map elements.
36340 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
36342 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
36343 Add linked-list forward and backlinks. Insert on
36344 construction, remove on destruction.
36345 (class pass_store_merging): Add m_stores_head field.
36346 (pass_store_merging::terminate_and_process_all_chains):
36347 Iterate over m_stores_head list.
36348 (pass_store_merging::terminate_all_aliasing_chains):
36350 (pass_store_merging::execute): Check for debug stmts first.
36351 Push new chains onto the m_stores_head stack.
36353 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
36356 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
36357 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
36358 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
36360 2017-03-16 Jeff Law <law@redhat.com>
36362 PR tree-optimization/71437
36363 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
36364 member function. Implementation moved into after_dom_children
36365 member function and into the threader's thread_outgoing_edges
36367 (dom_opt_dom_walker::after_dom_children): Simplify by moving
36368 some code into new thread_outgoing_edges.
36369 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
36370 definition. Simplify marker handling (do it here). Assume we always
36371 have the available expression and the const/copies tables.
36372 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
36374 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
36375 * tree-vrp.c (equiv_stack): No longer file scoped.
36376 (vrp_dom_walker): New class.
36377 (vrp_dom_walker::before_dom_children): New member function.
36378 (vrp_dom_walker::after_dom_children): Likewise.
36379 (identify_jump_threads): Setup domwalker. Use it rather than
36380 walking edges in a random order by hand. Simplify setup/finalization.
36381 (finalize_jump_threads): Remove.
36382 (vrp_finalize): Do not call identify_jump_threads here.
36383 (execute_vrp): Do it here instead and call thread_through_all_blocks
36386 PR tree-optimization/71437
36387 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
36389 (simplify_stmt_for_jump_threading): Add basic_block argument. All
36391 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
36392 (dom_opt_dom_walker::thread_across_edge): Remove
36393 handle_dominating_asserts argument. All callers changed.
36394 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
36395 changes. Remove calls to lhs_of_dominating_assert. Other
36396 uses of handle_dominating_asserts turn into unconditional code
36397 (simplify_control_stmt_condition_1): Likewise.
36398 (simplify_control_stmt_condition): Likewise.
36399 (thread_through_normal_block, thread_across_edge): Likewise.
36400 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
36401 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
36402 object if it is not an SSA_NAME.
36403 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
36404 before calling into the VRP specific simplifiers.
36405 (identify_jump_threads): Remove handle_dominating_asserts
36408 2017-03-16 Jakub Jelinek <jakub@redhat.com>
36411 * tree-diagnostic.c (default_tree_printer): No longer static.
36412 * tree-diagnostic.h (default_tree_printer): New prototype.
36414 2017-03-16 Tamar Christina <tamar.christina@arm.com>
36416 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
36417 Change ins into fmov.
36419 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36421 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
36422 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
36423 Use h_con constraint for operand 1.
36424 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
36425 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
36427 2017-03-15 Jeff Law <law@redhat.com>
36429 PR tree-optimization/71437
36430 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
36431 (record_temporary_equivalences): Use it.
36433 PR tree-optimization/71437
36434 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
36435 tree-ssa-scopedtables.
36436 (lookup_avail_expr, build_and_record_new_cond): Likewise.
36437 (record_conditions, record_cond, vuse_eq): Likewise.
36438 (record_edge_info): Adjust to API tweak of record_conditions.
36439 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
36440 (record_temporary_equivalences, optimize_stmt): Likewise.
36441 (eliminate_redundant_computations): Likewise.
36442 (record_equivalences_from_stmt): Likewise.
36443 * tree-ssa-scopedtables.c: Include options.h and params.h.
36444 (vuse_eq): New function, moved from tree-ssa-dom.c
36445 (build_and_record_new_cond): Likewise.
36446 (record_conditions): Likewise. Accept vector of conditions rather
36447 than edge_equivalence structure for first argument.
36448 for the first argument.
36449 (avail_exprs_stack::lookup_avail_expr): New member function, moved
36450 from tree-ssa-dom.c.
36451 (avail_exprs_stack::record_cond): Likewise.
36452 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
36453 from tree-ssa-dom.c.
36454 (avail_exprs_stack): Add new member functions lookup_avail_expr
36456 (record_conditions): Declare.
36458 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
36461 * lra-constraints.c (process_alt_operands): Increase reject for
36462 reloading an input/output operand.
36464 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
36467 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
36468 insns to convert from signed/unsigned char/short to IEEE 128-bit
36470 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
36472 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
36475 * config/i386/i386.c (ix86_vector_duplicate_value): Create
36476 subreg of inner mode for values already in registers.
36478 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
36480 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
36481 iteration reg is used after the loop.
36483 2017-03-14 Martin Sebor <msebor@redhat.com>
36485 PR tree-optimization/79800
36486 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
36487 precision in negative-positive range.
36488 (format_floating): Call non-const overload with adjusted precision.
36490 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
36493 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
36496 2017-03-14 Martin Sebor <msebor@redhat.com>
36498 PR middle-end/80020
36499 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
36500 * builtins.def (aligned_alloc): Use it.
36503 * Makefile.in (GTFILES): Add calls.c.
36504 * calls.c: Include "gt-calls.h".
36506 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
36508 PR rtl-optimization/79728
36509 * regs.h (struct target_regs): New field
36510 x_contains_allocatable_regs_of_mode.
36511 (contains_allocatable_regs_of_mode): New macro.
36512 * reginfo.c (init_reg_sets_1): Initialize it, and change
36513 contains_reg_of_mode so it includes global regs as well.
36514 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
36515 rather than contains_regs_of_mode.
36517 2017-03-14 Martin Liska <mliska@suse.cz>
36519 * doc/invoke.texi: Document options that can't be combined with
36520 -fcheck-pointer-bounds.
36522 2017-03-14 Martin Liska <mliska@suse.cz>
36524 PR middle-end/79831
36525 * doc/invoke.texi (-Wchkp): Document the option.
36527 2017-03-14 Martin Liska <mliska@suse.cz>
36529 * Makefile.in: Install gcov-dump.
36531 2017-03-14 Martin Liska <mliska@suse.cz>
36533 * multiple_target.c (expand_target_clones): Bail out for
36534 an invalid attribute.
36536 2017-03-14 Richard Biener <rguenther@suse.de>
36538 * alias.c (struct alias_set_entry): Pack properly.
36539 * cfgloop.h (struct loop): Likewise.
36540 * cse.c (struct set): Likewise.
36541 * ipa-utils.c (struct searchc_env): Likewise.
36542 * loop-invariant.c (struct invariant): Likewise.
36543 * lra-remat.c (struct cand): Likewise.
36544 * recog.c (struct change_t): Likewise.
36545 * rtl.h (struct address_info): Likewise.
36546 * symbol-summary.h (function_summary): Likewise.
36547 * tree-loop-distribution.c (struct partition): Likewise.
36548 * tree-object-size.c (struct object_size_info): Likewise.
36549 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
36550 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
36551 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
36552 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
36553 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
36554 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
36555 (struct _stmt_vec_info): Likewise.
36557 2017-03-14 Martin Liska <mliska@suse.cz>
36560 * multiple_target.c (create_dispatcher_calls): Check that
36561 a target can create a function dispatcher.
36563 2017-03-14 Martin Liska <mliska@suse.cz>
36566 * multiple_target.c (expand_target_clones): Drop local.local
36567 flag for default implementation.
36569 2017-03-14 Richard Biener <rguenther@suse.de>
36571 PR tree-optimization/80030
36572 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
36574 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
36576 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
36577 gcc_fallthrough() instead of __attribute__((fallthrough));
36579 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36581 * doc/gcc.texi: Remove "up" link to (DIR).
36582 * doc/gccint.texi: Ditto.
36584 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36586 * doc/install.texi (Specific) <avr>: Remove reference to
36589 2017-03-13 Jeff Law <law@redhat.com>
36591 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
36592 attribute rather than comments.
36594 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
36595 match_scratch operand is highest.
36597 2017-03-13 Martin Liska <mliska@suse.cz>
36599 PR middle-end/78339
36600 * ipa-pure-const.c (warn_function_noreturn): If the declarations
36601 is a CHKP clone, use original declaration.
36603 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36605 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
36606 (arc_conditional_register_usage): Use a different allocation order
36607 when optimizing for size.
36608 * common/config/arc/arc-common.c (arc_option_optimization_table):
36609 Section anchors default on when optimizing for size.
36611 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36613 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
36615 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36617 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
36618 * config/arc/arc.md (cpu_facility): Add cd variant.
36619 (*movqi_insn): Add code density variant.
36620 (*movhi_insn): Likewise.
36621 (*movqi_insn): Likewise.
36622 (*addsi3_mixed): Likewise.
36623 (subsi3_insn): Likewise.
36625 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36627 * config/arc/arc.md (movsi_cond_exec): Update constraint.
36629 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36631 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
36632 expressions with MINUS and UNARY ops.
36634 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36637 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
36639 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
36640 between vec_select and vector argument.
36641 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
36642 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
36643 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
36644 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
36645 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
36646 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
36648 2017-03-13 Richard Biener <rguenther@suse.de>
36651 * params.def (vect-max-peeling-for-alignment): Fix typo.
36653 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36655 * doc/install.texi (Specific) <mips-*-*>: Remove description of
36656 issue that only occurred with binutils below 2.18.
36658 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36660 * doc/install.texi (Specific) <cris-axis-elf>: No longer
36661 refer to binutils 2.11/2.12 minimum.
36663 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36665 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
36666 ftp.kernel.org and simplify binutils requirement.
36668 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
36670 * doc/invoke.texi (Warning Options): Fix spelling of link-time
36672 (Optimize Options): Ditto. Also remove redundancy.
36674 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36676 PR translation/79848
36677 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
36679 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
36680 to G_ to avoid double translation.
36682 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36684 PR translation/79923
36685 * auto-profile.c (get_combined_location): Convert leading
36686 character of diagnostics to lower case and remove trailing period.
36687 (read_profile): Likewise for various diagnostics.
36688 * config/arm/arm.c (arm_option_override): Remove trailing period
36689 from various diagnostics.
36690 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
36691 (msp430_expand_delay_cycles): Likewise.
36693 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36696 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
36697 full command-line argument, rather than just "str".
36698 (aarch64_validate_march): Likewise.
36699 (aarch64_validate_mtune): Likewise.
36701 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
36703 PR rtl-optimization/78911
36704 * lra-assigns.c (must_not_spill_p): New function.
36705 (spill_for): Use it.
36707 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36709 PR tree-optimization/79981
36710 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
36711 ATOMIC_COMPARE_EXCHANGE ifn result.
36712 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
36713 IFN_ATOMIC_COMPARE_EXCHANGE.
36715 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36718 * opts.c (parse_sanitizer_options): Add missing question mark to
36719 "did you mean" message.
36721 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36723 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
36725 (VMULEUH_UNS): Likewise.
36726 (VMULOUB_UNS): Likewise.
36727 (VMULOUH_UNS): Likewise.
36728 * config/rs6000/rs6000.c (builtin_function_type): Remove
36729 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
36731 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36734 * read-rtl-function.c (function_reader::read_rtx_operand): Update
36735 x with result of extra_parsing_for_operand_code_0.
36736 (function_reader::extra_parsing_for_operand_code_0): Convert
36737 return type from void to rtx, returning x. When reading
36738 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
36739 larger size containing struct block_symbol.
36741 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
36743 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
36744 -mfloat128-hardware without -m64.
36746 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
36749 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
36750 entries to the case statement that marks unsigned arguments to
36751 overloaded functions.
36753 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
36755 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
36756 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
36758 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
36761 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
36762 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
36764 2017-03-10 Martin Liska <mliska@suse.cz>
36768 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
36769 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
36772 2017-03-10 Olivier Hainque <hainque@adacore.com>
36774 * tree-switch-conversion (array_value_type): Start by resetting
36775 candidate type to it's main variant.
36777 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36779 PR rtl-optimization/79909
36780 * combine.c (try_combine): Use simplify_replace_rtx on individual
36781 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
36782 of the whole CALL_INSN_FUNCTION_USAGE.
36784 PR tree-optimization/79972
36785 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
36786 get_range_info on SSA_NAMEs. Formatting fixes.
36788 2017-03-10 Richard Biener <rguenther@suse.de>
36789 Jakub Jelinek <jakub@redhat.com>
36791 PR tree-optimization/77975
36792 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
36793 edge to be constant.
36794 (get_val_for): For constant x return it. Formatting fix.
36795 (loop_niter_by_eval): Avoid pointless looping if the next iteration
36796 would use the same bases as the current one.
36798 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36800 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
36801 instead of vec_select for V1TImode.
36802 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
36804 (VSX_LE_128): Add V1TI to this mode iterator.
36805 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36806 (*vsx_le_perm_store_<mode>): Likewise.
36807 (pre-reload splitter for VSX stores): Likewise.
36808 (post-reload splitter for VSX stores): Likewise.
36809 (*vsx_xxpermdi2_le_<mode>): Likewise.
36810 (*vsx_lxvd2x2_le_<mode>): Likewise.
36811 (*vsx_stxvd2x2_le_<mode>): Likewise.
36813 2017-03-09 Michael Eager <eager@eagercon.com>
36815 Correct failures with --enable-checking=yes,rtl.
36817 * config/microblaze/microblaze.c (microblaze_expand_shift):
36818 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36819 test for const0_rtx.
36820 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36821 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36823 2017-03-09 Richard Biener <rguenther@suse.de>
36825 PR tree-optimization/79977
36826 * graphite-scop-detection.c (scop_detection::merge_sese):
36827 Handle the case of extra exits to blocks dominating the entry.
36829 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36831 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36834 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36836 PR rtl-optimization/79949
36837 * lra-constraints.c (process_alt_operands): Check memory when
36838 trying to predict a cycle. Print about the overall increase.
36840 2017-03-09 Richard Biener <rguenther@suse.de>
36842 PR middle-end/79971
36843 * gimple-expr.c (useless_type_conversion_p): Preserve
36844 TYPE_SATURATING for fixed-point types.
36846 2017-03-09 Richard Biener <rguenther@suse.de>
36849 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36850 alignment of BLKmode params.
36852 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36855 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36856 (VALL_NO_V2Q): Likewise.
36858 * config/aarch64/aarch64-simd.md
36859 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36861 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36862 VALL_NO_V2Q mode iterator.
36863 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36865 2017-03-09 Martin Liska <mliska@suse.cz>
36867 PR tree-optimization/79631
36868 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36869 tree_int_cst_sign_bit just for INTEGER constants.
36871 2017-03-09 Martin Liska <mliska@suse.cz>
36875 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36878 2017-03-09 Marek Polacek <polacek@redhat.com>
36881 * tree.c (inchash::add_expr): Handle TREE_VEC.
36883 2017-03-09 Martin Liska <mliska@suse.cz>
36886 (chkp_narrow_size_and_offset): New function.
36887 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36888 (void chkp_parse_bit_field_ref): New function.
36889 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36890 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36892 2017-03-09 Martin Liska <mliska@suse.cz>
36895 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36896 (chkp_find_bounds_1): Remove gcc_unreachable.
36898 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36901 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36902 BUILT_IN_SYNC*, determine the access type from the size suffix and
36903 always build a MEM_REF with that type. Handle forgotten
36904 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36907 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36908 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36909 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36910 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36911 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
36912 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
36913 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
36914 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
36915 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
36916 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
36917 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
36918 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
36919 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
36920 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
36921 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
36922 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
36923 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
36924 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
36925 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
36926 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
36927 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
36928 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
36929 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
36930 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
36931 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
36932 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
36933 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
36934 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
36935 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
36936 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
36937 definitions outside of __OPTIMIZE__ guarded section.
36940 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
36941 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
36942 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
36943 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
36946 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36948 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
36949 ("vfenez<mode>"): Add missing constraints.
36951 2017-03-08 Martin Sebor <msebor@redhat.com>
36954 * config/nds32/nds32.c (nds32_option_override):
36955 Fix misspelled diagnostic.
36957 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36960 * gimplify.c (gimplify_omp_for): Replace index var in outer
36961 taskloop statement with an artificial variable and add
36962 OMP_CLAUSE_PRIVATE clause for it.
36964 2017-03-08 Richard Biener <rguenther@suse.de>
36966 PR tree-optimization/79955
36967 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
36968 for accesses that are completely outside of the variable.
36970 2017-03-08 Andrew Haley <aph@redhat.com>
36972 PR tree-optimization/79943
36973 * tree-ssa-loop-split.c (compute_new_first_bound): When
36974 calculating the new upper bound, (END-BEG) should be added, not
36977 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36979 * config/avr/avr.md (setmemhi): Make sure match_dup
36980 operand number comes before match_scratch.
36982 2017-03-08 Richard Biener <rguenther@suse.de>
36984 PR tree-optimization/79920
36985 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
36986 with ncopies == 1 to ...
36987 (vect_transform_slp_perm_load): ... here. Properly compute
36988 all element loads by iterating VF times over the group. Do
36989 not handle ncopies (computed in a broken way) in
36990 vect_create_mask_and_perm.
36992 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36995 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
36996 is a uniform vector, use uniform_vector_p return value instead of
36997 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
36999 2017-03-07 Marek Polacek <polacek@redhat.com>
37001 PR middle-end/79809
37002 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
37003 (alloca_call_type): Likewise.
37005 2017-03-07 Martin Liska <mliska@suse.cz>
37007 * gcov.c (process_args): Put comment to correct location.
37009 2017-03-07 Martin Liska <mliska@suse.cz>
37011 PR middle-end/68270
37012 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
37013 Use array_at_struct_end_p instead of DECL_CHAIN (field).
37014 (chkp_narrow_bounds_for_field): Likewise.
37015 (chkp_parse_array_and_component_ref): Pass one more argument to
37018 2017-03-07 Richard Biener <rguenther@suse.de>
37020 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
37023 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
37025 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
37026 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
37028 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37031 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
37032 to end of description.
37033 (PARAM_MAX_STORES_TO_MERGE): Likewise.
37035 2017-03-07 Jakub Jelinek <jakub@redhat.com>
37037 PR rtl-optimization/79901
37038 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
37040 (*avx512f_<code><mode>3<mask_name>): ... this.
37041 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
37042 iterator instead of VI8_AVX2_AVX512BW.
37044 PR rtl-optimization/79901
37045 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
37046 min/max expander, expand it using expand_vec_cond_expr.
37049 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
37052 2017-03-06 Jakub Jelinek <jakub@redhat.com>
37055 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
37056 to void * for PCH reasons.
37057 * dwarf2out.c (output_loc_operands, output_die): Cast
37058 v.val_vec.array to unsigned char *.
37060 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
37063 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
37066 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
37068 PR rtl-optimization/79571
37069 * lra-constraints.c (process_alt_operands): Calculate static
37070 reject and subtract it from overall when only addresses will be
37073 2017-03-06 Julia Koval <julia.koval@intel.com>
37076 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
37077 incoming stack boundary to 128 for 64-bit targets.
37079 2017-03-06 Richard Biener <rguenther@suse.de>
37081 PR tree-optimization/79894
37082 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
37083 to NULL after folding it.
37085 2017-03-06 Richard Biener <rguenther@suse.de>
37087 PR tree-optimization/79824
37088 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
37089 check disabling peeling for gaps.
37091 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
37093 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
37094 attributes): Document gettimeofday.
37096 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37098 * config/s390/s390.c (s390_option_override_internal): Set
37099 PARAM_MIN_VECT_LOOP_BOUND
37101 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37103 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
37104 * config/s390/s390.md: Likewise.
37106 2017-03-06 Jakub Jelinek <jakub@redhat.com>
37109 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
37110 (<avx2_avx512>_perm<mode>): Rename to ...
37111 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
37113 (<avx512>_perm<mode>_mask): Rename to ...
37114 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
37116 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
37117 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
37118 instead of VI8F_256_512.
37119 (avx512f_perm<mode>): New define_expand.
37120 (avx512f_perm<mode>_mask): Likewise.
37121 (avx512f_perm<mode>_1<mask_name>): New define_insn.
37122 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
37124 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37126 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
37127 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
37129 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
37131 2017-03-06 Martin Liska <mliska@suse.cz>
37134 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
37135 when having a SSA NAME w/o VAR_DECL assigned to it.
37137 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37139 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
37140 msa_dpsub_<su>_d): Fix MODE for vec_select.
37142 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37144 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
37146 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
37148 2017-03-06 Richard Biener <rguenther@suse.de>
37150 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
37151 * plugin.c (register_plugin_info): Likewise.
37152 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
37154 2017-03-05 Jakub Jelinek <jakub@redhat.com>
37156 * config/i386/sse.md (sse_storehps, sse_storelps,
37157 avx_<castmode><avxsizesuffix>_<castmode>,
37158 avx512f_<castmode><avxsizesuffix>_<castmode>,
37159 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
37160 in condition that at least one operand is not a MEM.
37162 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37164 PR middle-end/79805
37165 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
37166 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
37168 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
37169 gimple_call_nothrow_p flag based on whether original builtin can throw.
37170 If it can, emit following stmts on the fallthrough edge.
37171 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
37172 don't create new bb if inserting just debug stmts on the edge, try to
37173 insert them on the fallthru bb or just reset debug stmts.
37175 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
37178 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
37179 restore recog_data (including the operand rtxes inside it) around
37180 the call to get_insn_template.
37182 2017-03-03 Martin Sebor <msebor@redhat.com>
37184 PR tree-optimization/79699
37185 * context.c (context::~context): Free MPFR caches to avoid
37186 a memory leak on program exit.
37188 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37190 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
37191 Use wide_int::ulow () instead of .elt (0).
37193 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37195 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
37196 (*pushxf): Limit oF constraint to 32bit targets and add oC
37197 constraint for 64bit targets.
37198 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
37199 (*pushdf): Change rmF constraint to rmC.
37201 2017-03-03 Martin Liska <mliska@suse.cz>
37203 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
37204 Remove unused variable.
37206 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37209 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
37210 is a memory operand, increase num_memory.
37211 (ix86_expand_args_builtin): Likewise.
37213 2017-03-03 Jan Hubicka <jh@suse.cz>
37216 * ipa-devirt.c (maybe_record_node): Properly handle
37217 __cxa_pure_virtual visibility.
37219 2017-03-03 Martin Liska <mliska@suse.cz>
37221 PR tree-optimization/79803
37222 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
37224 (pass_loop_prefetch::execute): Disabled optimization if an
37225 assumption about L1 cache size is not met.
37227 2017-03-03 Martin Liska <mliska@suse.cz>
37229 PR rtl-optimization/79574
37230 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
37231 (hash_scan_set): Likewise.
37232 (dump_hash_table): Likewise.
37233 (hoist_code): Likewise.
37235 2017-03-03 Richard Biener <rguenther@suse.de>
37237 * fixed-value.c (fixed_from_string): Restore use of elt (1)
37238 in place of uhigh ().
37239 (fixed_convert_from_real): Likewise.
37241 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37244 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
37246 2017-03-03 Richard Biener <rguenther@suse.de>
37248 PR middle-end/79818
37249 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
37250 TYPE_OVERFLOW_UNDEFINED check.
37252 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37254 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
37256 (vector_ae_<mode>_p): Likewise.
37257 (vector_nez_<mode>_p): Likewise.
37258 (vector_ne_v2di_p): Likewise.
37259 (vector_ae_v2di_p): Likewise.
37260 (vector_ne_<mode>_p): Likewise.
37261 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
37263 (vsx_tsqrt<mode>2_fe): Likewise.
37265 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
37268 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
37270 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37272 PR rtl-optimization/79780
37273 * cprop.c (one_cprop_pass): When second and further conditional trap
37274 in a single basic block is turned into an unconditional trap, turn it
37275 into a deleted note to avoid RTL verification failures.
37277 2017-03-02 Richard Biener <rguenther@suse.de>
37279 * fold-const.c (const_binop): Use ulow () instead of elt (0).
37281 2017-03-02 Richard Biener <rguenther@suse.de>
37283 PR tree-optimization/79345
37285 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
37286 param and abort the walk, returning -1 if it is hit.
37287 (walk_aliased_vdefs): Take a limit param and pass it on.
37288 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
37289 defaulting to 0 and return a signed int.
37290 * tree-ssa-uninit.c (struct check_defs_data): New struct.
37291 (check_defs): New helper.
37292 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
37293 about uninitialized memory.
37294 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
37295 bogus uninitialized warning.
37296 (fixed_convert_from_real): Likewise.
37298 2017-03-02 Bin Cheng <bin.cheng@arm.com>
37300 PR tree-optimization/66768
37301 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
37302 iv_use if base object can't be determined.
37304 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37306 PR tree-optimization/79345
37307 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
37308 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
37309 (get_pattern_stats): Initialize it.
37310 * genemit.c (gen_expand): Verify match_scratch numbers come after
37311 match_operand/match_dup numbers.
37312 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
37313 match_scratch numbers.
37314 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
37316 * config/s390/s390.md (trunctdsd2): Likewise.
37318 2017-03-02 Richard Biener <rguenther@suse.de>
37320 * wide-int.h (wide_int_storage::operator=): Implement in terms
37323 2017-03-02 Richard Biener <rguenther@suse.de>
37325 PR tree-optimization/79777
37326 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
37327 the to insert expression to sth existing.
37329 2017-03-01 Martin Sebor <msebor@redhat.com>
37331 PR middle-end/79692
37332 * gimple-ssa-sprintf.c
37333 (directive::known_width_and_precision): New function.
37334 (format_integer): Use it.
37335 (get_mpfr_format_length): Consider the full range of precision
37336 when computing %g output with the # flag. Set the likely byte
37337 count to 3 rather than 1 when precision is indeterminate.
37338 (format_floating): Correct the lower bound of precision.
37340 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37342 * doc/invoke.texi: Document default code model for 64-bit Linux.
37344 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37347 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
37348 udiv rather than div since input pattern is unsigned.
37350 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
37352 * config/i386/i386.c (print_reg): Warn for values of
37353 unsupported size in integer register.
37355 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
37358 * config/rs6000/predicates.md (current_file_function_operand): Do
37359 not allow self calls to be local if the function is replaceable.
37361 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
37364 * config/rs6000/altivec.h (vec_ctz and others): Change the
37365 preprocessor macro that controls conditional compilation from
37366 _ARCH_PWR9 to __POWER9_VECTOR__.
37367 (vec_all_ne): Change parameterization of __altivec_scalar_pred
37368 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
37369 control (instead of _ARCH_PWR9 control) so that template
37370 definition uses power9-specific function.
37371 (vec_any_eq): Likewise.
37372 (vec_all_ne): Change macro definition to use a power9-specific
37373 expansion under #ifdef __POWER9_VECTOR__ control (instead of
37374 _ARCH_PWR9 control).
37375 (vec_any_eq) Likewise.
37376 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
37377 expansion for CMPNEF to remove support for xvcmpnesp instruction.
37378 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
37379 support for xvcmpnedp instruction.
37380 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
37381 macro expansion so that Power9 implementation of vec_all_ne does
37382 not use the AltiVec predicate framework.
37383 (VCMPNEH_P): Likewise.
37384 (VCMPNEW_P): Likewise.
37385 (VCMPNED_P): Likewise.
37386 (VCMPNEFP_P): Likewise.
37387 (VCMPNEDP_P): Likewise.
37388 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
37389 implementation of vec_any_eq to not use AltiVec predicate
37391 (VCMPAEH_P): Likewise.
37392 (VCMPAEW_P): Likewise.
37393 (VCMPAED_P): Likewise.
37394 (VCMPAEFP_P): Likewise.
37395 (VCMPAEDP_P): Likewise.
37396 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
37397 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
37398 not use the AltiVec predicate framework.
37399 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
37400 of vec_any_eq to not use AltiVec predicate framework.
37401 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
37402 support for predefined __POWER9_VECTOR__ macro to indicate that
37403 Power9 instruction selection is enabled.
37404 (altivec_overloaded_builtins): Remove extraneous
37405 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
37406 function argument types RS6000_BTI_bool_V16QI and
37407 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
37408 entry for overloaded function argument types RS6000_BTI_bool_V4SI
37409 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
37410 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
37411 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
37412 Power9 for implementations of vec_cmpne. Change the signature for
37413 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
37414 (representing vec_all_ne) to remove the previously described first
37415 argument of type RS6000_BTI_INTSI, as this was an artifact of
37416 reliance on the AltiVec predicate framework, which is no longer
37417 used in the implementation of these functions. Add
37418 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
37419 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
37420 since, unlike the AltiVec predicate framework implementation, we
37421 do not share function descriptors between vec_alle and vec_anyeq.
37422 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
37423 set of modes that receive special treatment even when
37424 TARGET_P9_VECTOR is true. The special treatment emits code that
37425 does not depend on Power9 instructions.
37426 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
37427 define_expand to not rely on AltiVec predicate framework.
37428 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37430 (vector_ne_v2di_p): Change this define_expand to not rely on
37431 AltiVec predicate framework.
37432 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
37434 (vector_ne_<mode>_p): Change this define_expand to not rely on
37435 AltiVec predicate framework.
37436 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37438 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
37439 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
37440 define_insn pattern.
37441 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
37442 define_insn pattern because the xvcmpne<VSs>. instruction is not
37444 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
37445 instruction is not supported.
37447 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37449 * config/nvptx/nvptx.c: Include intl.h.
37451 2017-03-01 Martin Jambor <mjambor@suse.cz>
37454 * ipa-prop.h (ipa_bits): Removed field known.
37455 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
37456 to pointers. Adjusted their comments to warn about their sharing.
37457 (ipcp_transformation_summary): Change bits to a vector of pointers.
37458 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
37459 (ipa_get_ipa_bits_for_value): Declare.
37460 * tree-vrp.h (value_range): Mark as GTY((for_user)).
37461 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
37462 (ipa_bits_hash_table): Likewise.
37463 (ipa_vr_ggc_hash_traits): Likewise.
37464 (ipa_vr_hash_table): Likewise.
37465 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
37466 being pointers and vr_known being removed.
37467 (ipa_set_jf_unknown): Likewise.
37468 (ipa_get_ipa_bits_for_value): New function.
37469 (ipa_set_jfunc_bits): Likewise.
37470 (ipa_get_value_range): New overloaded functions.
37471 (ipa_set_jfunc_vr): Likewise.
37472 (ipa_compute_jump_functions_for_edge): Use the above functions to
37473 construct bits and vr parts of jump functions.
37474 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
37475 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37477 (ipcp_grow_transformations_if_necessary): Also allocate
37478 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37480 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
37481 them. Fix too long lines.
37482 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
37483 vr_known being removed.
37484 (ipa_read_jump_function): Use new setter functions to construct bits
37485 and vr parts of jump functions or set them to NULL.
37486 (write_ipcp_transformation_info): Adjust for bits being pointers.
37487 (read_ipcp_transformation_info): Likewise.
37488 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
37490 Include gt-ipa-prop.h.
37491 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
37493 (ipcp_store_bits_results): Likewise.
37494 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
37495 Do not write to existing jump functions but use a temporary instead.
37497 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37500 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
37501 attempt to use its first operand as BIT_FIELD_REF base.
37503 2017-03-01 Richard Biener <rguenther@suse.de>
37505 PR middle-end/79721
37506 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
37507 interpolating formula in wrapping arithmetic.
37508 (chrec_apply): Convert chrec_evaluate return value to wanted type.
37510 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37512 PR tree-optimization/79734
37513 * tree-vect-generic.c (expand_vector_condition): Optimize
37514 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
37515 Handle VEC_COND_EXPR where comparison has different inner width from
37516 type's inner width.
37518 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
37520 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
37521 markup, and similar issues. Remove @opindex entries for things
37522 that aren't options. Add missing -mmpy-option entries.
37524 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37526 PR tree-optimization/79737
37527 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
37528 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
37529 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
37530 instead of byte_size. Formatting fix.
37531 (shift_bytes_in_array_right): Formatting fix.
37533 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
37536 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
37537 condition on optimize for the leaf function test.
37539 2017-02-28 Martin Liska <mliska@suse.cz>
37542 * read-rtl-function.c (function_reader::handle_unknown_directive):
37543 Bail out when one uses -flto.
37545 2017-02-28 Martin Liska <mliska@suse.cz>
37547 * common.opt: Replace space with tabular for options of <number>
37549 * config/i386/i386.opt: Show <number> value for
37550 -mlarge-data-threshold.
37551 * opts.c (print_filtered_help): Do not display number in hexadecimal
37554 2017-02-28 Martin Liska <mliska@suse.cz>
37556 * common.opt: Fix --help=option -Q for options which are of
37559 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
37561 * config/i386/i386.c (print_reg): Error out for values
37562 of 8-bit size in invalid integer register.
37564 2017-02-28 Martin Sebor <msebor@redhat.com>
37566 PR tree-optimization/79691
37567 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
37569 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37572 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
37573 gcc_unreachable with output_operand_lossage.
37575 2017-02-28 Richard Biener <rguenther@suse.de>
37577 PR tree-optimization/79740
37578 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
37580 (visit_nary_op): Insert the nary into the hashtable if we
37581 pattern-matched sth.
37582 * tree-ssa-pre.c (eliminate_insert): Robustify.
37584 2017-02-28 Richard Biener <rguenther@suse.de>
37586 PR middle-end/79731
37587 * fold-const.c (decode_field_reference): Reject out-of-bound
37590 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37592 * config/i386/i386.c: Include intl.h.
37593 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
37594 instead of just cond ? "..." : "...".
37595 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
37596 * coverage.c (read_counts_file): Likewise.
37597 * omp-offload.c: Include intl.h.
37598 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
37599 of just cond ? "..." : "...".
37600 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
37601 of just cond ? "..." : "...".
37603 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
37606 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
37608 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
37609 'tune for' CPU name.
37610 * config/arm/arm-cpu-data.h: Regenerated.
37612 2017-02-28 Richard Biener <rguenther@suse.de>
37614 PR tree-optimization/79732
37615 * tree-inline.c (expand_call_inline): Do not shadow var.
37617 2017-02-28 Richard Biener <rguenther@suse.de>
37619 PR tree-optimization/79723
37620 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
37621 address-space properly.
37623 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
37625 * doc/optinfo.texi (Optimization groups): Fix option used for
37627 * doc/invoke.texi (-fopt-info): Document "omp".
37628 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
37629 (OPTGROUP_ALL): Add OPTGROUP_OMP.
37630 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
37631 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
37632 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
37634 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
37636 * dumpfile.c (optgroup_options): Instead of "openmp", associate
37637 OPTGROUP_OMP with "omp".
37639 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
37642 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
37643 for arithmetic shift of unsigned V2DI.
37645 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
37647 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
37648 arc/linux.h headers.
37649 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
37650 (LINK_SPEC): Likewise.
37651 (ARC_TLS_EXTRA_START_SPEC): Likewise.
37652 (EXTRA_SPECS): Likewise.
37653 (STARTFILE_SPEC): Likewise.
37654 (ENDFILE_SPEC): Likewise.
37655 (LIB_SPEC): Likewise.
37656 (TARGET_SDATA_DEFAULT): Likewise.
37657 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
37658 (MULTILIB_DEFAULTS): Likewise.
37659 (DWARF2_UNWIND_INFO): Likewise.
37660 * config/arc/big.h: New file.
37661 * config/arc/elf.h: Likewise.
37662 * config/arc/linux.h: Likewise.
37663 * config/arc/t-uClibc: Remove.
37665 2017-02-27 Bin Cheng <bin.cheng@arm.com>
37667 PR tree-optimization/77536
37668 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
37669 (tree_transform_and_unroll_loop): Use above function to compute the
37670 estimated niter of unrolled loop and use it when scaling profile.
37671 Also use count info rather than frequency if it's non-zero.
37672 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
37673 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
37674 (vect_transform_loop): Call above function.
37676 2017-02-27 Richard Biener <rguenther@suse.de>
37678 PR tree-optimization/45397
37679 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
37680 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
37681 (visit_nary_op): Add pattern matching for CSEing sign-changed
37682 or truncated operations with wider ones.
37684 2017-02-27 Richard Biener <rguenther@suse.de>
37686 PR tree-optimization/79690
37687 * tree-vect-stmts.c (vectorizable_store): Use vector type
37688 built from the DR with address-space.
37690 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
37692 * doc/invoke.texi (Optimize Options): Refine the description
37693 of asan-use-after-return.
37695 2017-02-25 Alan Modra <amodra@gmail.com>
37697 PR rtl-optimization/79584
37698 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
37699 base, not ad->base_term, the reg within base. Remove assertion
37700 that ad->base == ad->base_term. Replace gen_int_mode using
37701 bogus mode with const0_rtx.
37703 2017-02-25 Jakub Jelinek <jakub@redhat.com>
37705 PR middle-end/79396
37706 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
37707 FMA_EXPR like tcc_binary or tcc_unary.
37709 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
37712 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
37714 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
37715 (output_loc_operands): Handle DW_OP_call_ref and
37716 DW_OP_GNU_variable_value.
37717 (struct variable_value_struct): New type.
37718 (struct variable_value_hasher): Likewise.
37719 (variable_value_hash): New variable.
37720 (string_types): Remove.
37721 (copy_loc_descr): New function.
37722 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
37723 (prepend_loc_descr_to_each): New function.
37724 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
37725 instead of add_loc_descr_to_each if the first argument is single
37726 location list and the second has multiple.
37727 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
37728 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
37729 when looking for variable value which doesn't have other location info.
37730 (loc_list_from_tree): Formatting fix.
37731 (gen_array_type_die): Simplify DW_AT_string_length handling.
37732 (adjust_string_types): Remove.
37733 (gen_subprogram_die): Don't call adjust_string_types nor test/set
37734 string_types. Call resolve_variable_values.
37735 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
37736 (resolve_addr_in_expr): Likewise. Add A argument.
37737 (copy_deref_exprloc): Remove deref argument. Adjust for the
37738 original expression being DW_OP_GNU_variable_value with optionally
37739 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
37740 optionally after it.
37741 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
37742 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
37743 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
37744 (variable_value_hasher::hash, variable_value_hasher::equal): New
37746 (resolve_variable_value_in_expr, resolve_variable_value,
37747 resolve_variable_values, note_variable_value_in_expr,
37748 note_variable_value): New functions.
37749 (dwarf2out_early_finish): Call note_variable_value on all toplevel
37752 2017-02-24 Jakub Jelinek <jakub@redhat.com>
37755 * opts.h (handle_generated_option): Add GENERATED_P argument.
37756 * opts-common.c (handle_option): Adjust function comment.
37757 (handle_generated_option): Add GENERATED_P argument, pass it to
37759 (control_warning_option): Pass false to handle_generated_option
37761 * opts.c (maybe_default_option): Pass true to handle_generated_option
37763 * optc-gen.awk: Likewise.
37765 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37767 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
37768 a REG, look at the REG it is a SUBREG of.
37769 (splitter for cmpeqsi_t): Ditto.
37771 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37773 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
37774 the special USEs with the pattern of the insn, not the insn itself.
37776 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
37779 * doc/invoke.texi: Document -mload-store-pairs.
37781 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37782 Sandra Loosemore <sandra@codesourcery.com>
37784 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
37785 argument isn't a CONST_INT.
37786 (nios2_alternate_compare_const): Assert op is a CONST_INT.
37787 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
37788 (nios2_validate_compare): Bypass alternate compare logic if *op2
37789 is not a CONST_INT.
37790 (ldstwm_operation_p): Return false if first_base is not a REG or
37791 if first_offset is not a CONST_INT.
37793 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37795 * config/cris/cris.md: Use correct operand in a define_peephole2.
37797 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37799 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
37801 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37803 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
37804 this_insn if it is an INSN or JUMP_INSN.
37805 (force_offsettable): Look at base, not at addr.
37806 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37807 on things that aren't necessarily CONST_INTs.
37809 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37811 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37812 -mfpmath=sse is the default also for x86-32 targets with SSE2
37813 instruction set when @option{-ffast-math} is enabled
37815 2017-02-24 Jeff Law <law@redhat.com>
37817 PR rtl-optimizatoin/79286
37818 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37821 2017-02-24 Richard Biener <rguenther@suse.de>
37823 PR tree-optimization/79389
37824 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37827 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37829 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37830 function comment to reflect reality.
37831 (loop_exits_before_overflow): Fix typo in function description.
37833 2017-02-24 Richard Biener <rguenther@suse.de>
37835 PR tree-optimization/79389
37836 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37837 properly that a threading opportunity exists. Detect conditional
37838 copy/constant propagation opportunities.
37840 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37842 * config/visium/visium.md (type): Add trap.
37843 (b): New mode attribute.
37844 (*btst): Rename into...
37845 (*btst<mode>): ...this and adjust.
37846 (*cbranchsi4_btst_insn): Rename into...
37847 (*cbranch<mode>4_btst_insn): ...this and adjust.
37848 (trap): New define_insn.
37850 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37852 PR tree-optimization/79389
37853 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37854 (noce_reversed_cond_code): New function.
37855 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37856 reversed_comparison_code. Formatting fix.
37857 (noce_try_store_flag): Test rev_cond != NULL in addition to
37858 reversed_comparison_code.
37859 (noce_try_store_flag_constants): Likewise.
37860 (noce_try_store_flag_mask): Likewise.
37861 (noce_try_addcc): Use rev_cond if non-NULL instead of
37862 reversed_comparison_code.
37863 (noce_try_cmove_arith): Likewise. Formatting fixes.
37864 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37865 (noce_find_if_block): Initialize rev_cond.
37866 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37867 instead of false as last argument never attempt to reverse it
37870 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37872 PR tree-optimization/79663
37873 * tree-predcom.c (combine_chains): Process refs in reverse order
37874 only for ZERO length chains, and add explaining comment.
37876 2017-02-23 Jeff Law <law@redhat.com>
37878 PR tree-optimization/79578
37879 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37880 in call to operand_equal_p.
37882 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37885 * config/i386/cpuid.h: Fix another undefined behavior.
37887 2017-02-23 Richard Biener <rguenther@suse.de>
37889 PR tree-optimization/79683
37890 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37891 vector types for data-refs.
37893 2017-02-23 Martin Liska <mliska@suse.cz>
37895 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37897 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37899 PR middle-end/79665
37900 * internal-fn.c (get_range_pos_neg): Moved to ...
37901 * tree.c (get_range_pos_neg): ... here. No longer static.
37902 * tree.h (get_range_pos_neg): New prototype.
37903 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37904 are known to be in between 0 and signed maximum inclusive, try to
37905 expand both unsigned and signed divmod and use the cheaper one from
37908 2017-02-22 Jeff Law <law@redhat.com>
37910 PR tree-optimization/79578
37911 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37912 to compare base operands.
37914 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
37917 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
37918 gpc_reg_operand instead of fpr_reg_operand.
37920 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
37922 * config/mips/mips.c (mips_return_in_memory): Force FP
37923 vector types to be returned in memory for o32 ABI.
37925 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37927 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
37928 instead of DW_TAG_member for static data member declarations and don't
37929 set no_linkage_name for static inline data members.
37930 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
37933 2017-02-22 Martin Liska <mliska@suse.cz>
37935 * doc/invoke.texi: Replace inequality signs with square brackets
37938 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
37941 * lra-constraints.c (simplify_operand_subreg): Handle
37942 WORD_REGISTER_OPERATIONS targets.
37944 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37947 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
37948 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
37949 elimination by swapping fld*.
37951 2017-02-22 Richard Biener <rguenther@suse.de>
37953 PR tree-optimization/79673
37954 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
37955 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
37956 irrelevant address-space qualifiers and avoiding a
37957 ADDR_SPACE_CONVERT_EXPR from fold_convert.
37959 2017-02-22 Richard Biener <rguenther@suse.de>
37961 PR tree-optimization/79666
37962 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
37963 to not symbolically negate if that may introduce undefined
37966 2017-02-22 Martin Liska <mliska@suse.cz>
37969 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
37970 * data-streamer-out.c (streamer_write_gcov_count_stream):
37972 * value-prof.c (stream_out_histogram_value): Make assert more
37973 precise based on type of counter.
37975 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
37978 * config/i386/i386.md (standard_x87sse_constant_load splitter):
37979 Use nonimmediate_operand instead of memory_operand for operand 1.
37980 (float-extend standard_x87sse_constant_load splitter): Ditto.
37982 2017-02-21 Jeff Law <law@redhat.com>
37984 PR tree-optimization/79621
37985 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
37986 blocks with edges to themselves.
37988 2017-02-21 Jakub Jelinek <jakub@redhat.com>
37991 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
37992 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
37993 Use gimple_call_builtin_p.
37996 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
37997 on temporarily removed DEBUG_INSNs.
37999 PR tree-optimization/79649
38000 * tree-loop-distribution.c (classify_partition): Give up on
38001 non-generic address space loads/stores.
38003 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
38005 * doc/loop.texi (Loop manipulation): Remove nonexistent
38006 tree_ssa_loop_version from the documentation.
38007 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
38009 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38012 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
38013 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
38014 * config/rs6000/rs6000.c: Include except.h.
38015 (rs6000_expand_split_stack_prologue): Call
38016 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
38018 2017-02-21 Martin Jambor <mjambor@suse.cz>
38021 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
38022 have been analyzed.
38024 2017-02-21 Martin Jambor <mjambor@suse.cz>
38026 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
38027 for backward compatibility only.
38028 * doc/invoke.texi (Option Summary): Remove all references to
38029 -fipa-cp-alignment.
38031 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
38035 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38037 * lra-constraints.c (curr_insn_transform): Handle
38038 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
38040 2017-02-21 Martin Liska <mliska@suse.cz>
38042 * config/i386/i386.opt: Replace -masm-dialect with -masm.
38044 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
38046 PR translation/79638
38047 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
38049 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
38052 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
38053 (arm_function_ok_for_sibcall): Return false for an indirect call by
38054 descriptor if all the argument registers are used.
38055 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
38056 alignment of the function.
38058 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38060 PR tree-optimization/61441
38061 * simplify-rtx.c (simplify_const_unary_operation): For
38062 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
38063 the sNaN unmodified.
38065 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
38067 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
38068 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
38069 instead of SYSTEM_HEADER_DIR.
38071 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
38072 Martin Liška <mliska@suse.cz>
38074 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
38075 Fix typos and grammar, use active voice, and clarify.
38077 2017-02-20 Marek Polacek <polacek@redhat.com>
38079 PR middle-end/79537
38080 * gimplify.c (gimplify_expr): Handle unused *&&L;.
38083 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
38085 2017-02-20 Jakub Jelinek <jakub@redhat.com>
38088 * config/i386/i386.c (ix86_expand_builtin): Handle
38089 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
38090 ix86_builtins_isa[fcode].isa as a requirement of those
38091 flags and any other flag in the bitmask.
38092 (ix86_init_mmx_sse_builtins): Use 0 instead of
38093 ~OPTION_MASK_ISA_64BIT as mask.
38094 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
38095 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
38096 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
38097 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
38099 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38102 * lra-constraints.c (split_reg): Check requested split mode
38103 is supported by the register.
38105 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38107 * lra-constraints.c (simplify_operand_subreg): Remove early
38110 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38113 * lra-constraints.c (curr_insn_transform): Tighten condition
38114 for converting SUBREG reloads from OP_OUT to OP_INOUT.
38116 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38119 * lra-constraints.c (curr_insn_transform): Handle
38120 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
38122 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
38125 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
38127 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
38129 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
38132 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
38135 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38137 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
38140 2017-02-18 Jakub Jelinek <jakub@redhat.com>
38142 * final.c (last_columnnum, override_columnnum): New variables.
38143 (final_start_function): Set last_columnnum, pass it to begin_prologue
38144 hook and pass 0 to dwarf2out_begin_prologue.
38145 (final_scan_insn): Update override_columnnum. Pass last_columnnum
38146 to source_line debug hook.
38147 (notice_source_line): Compute last_columnnum and for debug_column_info
38148 return true on column changes.
38149 * debug.h (struct gcc_debug_hooks): Add column argument to
38150 source_line and begin_prologue hooks.
38151 (debug_nothing_int_charstar_int_bool): Remove prototype.
38152 (debug_nothing_int_int_charstar,
38153 debug_nothing_int_int_charstar_int_bool): New prototypes.
38154 (dwarf2out_begin_prologue): Add column argument.
38155 * debug.c (do_nothing_debug_hooks): Adjust source_line and
38156 begin_prologue hooks.
38157 (debug_nothing_int_charstar_int_bool): Remove.
38158 (debug_nothing_int_int_charstar,
38159 debug_nothing_int_int_charstar_int_bool): New functions.
38160 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
38161 through to dwarf2out_source_line.
38162 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
38163 (dwarf2out_source_line): Add column argument, emit it if requested.
38164 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
38166 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38167 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38168 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
38169 through to dwarf2out_begin_prologue.
38170 (vmsdbgout_source_line): Add column argument, pass it through to
38171 dwarf2out_source_line.
38172 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
38173 dbxout_source_line caller.
38174 (dbxout_source_line): Add column argument.
38176 * common.opt (gno-column-info, gcolumn-info): New options.
38177 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
38178 (check_die): Also test for multiple DW_AT_decl_column attributes.
38179 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
38180 DW_AT_decl_column if requested.
38181 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
38183 (gen_variable_die): Likewise.
38184 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
38185 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
38188 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
38189 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
38190 (ix86_handle_option): Handle OPT_m3dnowa.
38191 * doc/invoke.texi (-m3dnowa): Document.
38192 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
38193 -m3dnowa instead of -m3dnow -march=athlon.
38196 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
38197 instead of gcc_assert for K, r and R code checks. Formatting fixes.
38199 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38202 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
38203 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
38204 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
38205 generator for vsx_xxpermdi_<mode>_be.
38206 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
38207 force big-endian semantics.
38208 (vsx_xxpermdi_<mode>_be): New define_expand with same
38209 implementation as previous version of vsx_xxpermdi_<mode>.
38211 2017-02-17 Jakub Jelinek <jakub@redhat.com>
38213 PR tree-optimization/79327
38214 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
38215 variable, its initialization and use.
38217 2017-02-17 Julia Koval <julia.koval@intel.com>
38219 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
38220 (OPTION_MASK_ISA_PKU_UNSET): New.
38221 (ix86_handle_option): Handle -mrdpid.
38222 * config/i386/cpuid.h (bit_RDPID): New.
38223 * config/i386/driver-i386.c (host_detect_local_cpu):
38224 Detect RDPID feature.
38225 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
38226 * config/i386/i386-c.c (ix86_target_macros_internal):
38228 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
38229 (ix86_valid_target_attribute_inner_p): Add "rdpid".
38230 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
38231 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
38232 * config/i386/i386.md (define_insn "rdpid"): New.
38233 * config/i386/i386.opt Add -mrdpid.
38234 * config/i386/immintrin.h (_rdpid_u32): New.
38236 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
38238 PR rtl-optimization/79541
38239 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
38240 instead of transforming it into USE.
38242 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
38244 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
38245 If HONOR_SNANS (SFmode) force the input to a register.
38246 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
38247 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
38248 an frsp or similar insn.
38250 2017-02-17 Martin Liska <mliska@suse.cz>
38252 PR rtl-optimization/79577
38253 * params.def (selsched-max-sched-times): Increase minimum to 1.
38255 2017-02-17 Martin Liska <mliska@suse.cz>
38257 PR rtl-optimization/79574
38258 * gcse.c (want_to_gcse_p): Prevent integer overflow.
38260 2017-02-17 Martin Liska <mliska@suse.cz>
38262 PR tree-optimization/79529
38263 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
38264 ssa_defined_default_def_p to handle cases which are implicitly
38266 * tree-ssa.c (ssa_defined_default_def_p): New function.
38267 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
38268 which are implicitly defined.
38269 * tree-ssa.h (ssa_defined_default_def_p): Declare.
38271 2017-02-17 Richard Biener <rguenther@suse.de>
38273 PR middle-end/79576
38274 * params.def (max-ssa-name-query-depth): Limit to 10.
38276 2017-02-17 Richard Biener <rguenther@suse.de>
38278 PR tree-optimization/79552
38279 * tree-ssa-structalias.c (visit_loadstore): Properly verify
38282 2017-02-17 Richard Biener <rguenther@suse.de>
38285 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
38287 2017-02-17 Marek Polacek <polacek@redhat.com>
38289 PR middle-end/79536
38290 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
38291 (fold_negate_expr): New wrapper.
38293 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
38295 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
38296 Correct terminology and de-emphasize pre-standard behavior.
38298 2017-02-16 Alan Modra <amodra@gmail.com>
38300 PR rtl-optimization/79286
38301 * ira.c (def_dominates_uses): New function.
38302 (update_equiv_regs): Don't create an equivalence for insns that
38303 may trap where the register def does not dominate the use.
38305 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
38307 PR rtl-optimization/78127
38308 * lra.c (lra): Call lra_eliminate before finish the loop after
38311 2017-02-16 Richard Biener <rguenther@suse.de>
38313 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
38315 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
38316 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
38317 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
38318 (isl_val_int_from_wi): New function.
38319 (extract_affine_gmp): Rename to ...
38320 (extract_affine_wi): ... this, take a widest_int.
38321 (extract_affine_int): Just wrap extract_affine_wi.
38322 (add_param_constraints): Use isl_val_int_from_wi.
38323 (add_loop_constraints): Likewise, and extract_affine_wi.
38325 2017-02-15 Jeff Law <law@redhat.com>
38327 PR middle-end/79521
38328 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
38329 ira_init_register_move_cost_if_necessary.
38331 2017-02-15 Martin Sebor <msebor@redhat.com>
38333 PR middle-end/32003
38334 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
38335 removed in a prior commit.
38337 2017-02-15 Bin Cheng <bin.cheng@arm.com>
38339 PR tree-optimization/79347
38340 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
38341 counters during peeling.
38343 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
38345 * Makefile.in (site.exp): Remove "set ISLVER".
38347 2017-02-15 Jakub Jelinek <jakub@redhat.com>
38350 * real.c (real_from_integer): Call real_convert even for decimal.
38352 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
38355 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
38357 2017-02-14 Andrew Pinski <apinski@cavium.com>
38359 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
38360 cores and change the partno/implementer to be correct.
38361 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
38362 the 'B" as the implementer.
38363 * config/aarch64/aarch64-tune.md: Regenerate.
38365 2017-02-14 Carl Love <cel@us.ibm.com>
38367 * config/rs6000/rs6000.c: Add case statement entry to make the
38368 xvcvuxdsp built-in argument unsigned.
38369 * config/rs6000/vsx.md: Fix the source and return operand types so they
38370 match the instruction definitions from the ISA document. Fix typo
38371 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
38374 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
38377 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
38378 member early_clobber_alts.
38379 * lra-lives.c (reg_early_clobber_p): New.
38380 (process_bb_lives): Use it.
38381 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
38382 (debug_operand_data): Initialize early_clobber_alts.
38383 (setup_operand_alternative): Set up early_clobber_alts.
38384 (collect_non_operand_hard_regs): Ditto. Pass early clobber
38385 alternatives to new_insn_reg.
38386 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
38388 (lra_update_insn_regno_info): Pass the new arg.
38390 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38392 PR middle-end/79505
38393 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
38394 (new_oacc_loop_raw): Don't clear already cleared fields.
38397 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
38398 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
38399 _mm512_prefetch_i64gather_ps): New inline functions and macros.
38401 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
38404 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
38406 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
38409 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
38410 the extra instruction to the right place to store 128-bit constant
38413 2017-02-14 Martin Sebor <msebor@redhat.com>
38415 PR middle-end/79448
38416 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
38417 warning for strings of unknown length.
38419 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
38421 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
38423 2017-02-14 Jeff Law <law@redhat.com>
38426 * ira-costs.c (scan_one_insn): Initialize register move costs
38427 for pseudos seen in USE/CLOBBER insns.
38429 PR tree-optimization/79095
38430 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
38431 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
38432 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
38433 if the operands are known to be not equal, then the resulting range
38435 (intersect_ranges): If the new range is ~[0,0] and the old range is
38436 wide, then prefer ~[0,0].
38437 * tree-vrp.c (overflow_comparison_p_1): New function.
38438 (overflow_comparison_p): New function.
38439 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
38440 if NAME is used in an overflow test.
38441 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
38442 overflow check that can be expressed as an equality test, then adjust
38443 ops to be that equality test.
38445 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38447 * config/s390/s390-builtin-types.def: Remove flags argument.
38448 * config/s390/s390.c (s390_init_builtins): Likewise.
38450 2017-02-14 Martin Liska <mliska@suse.cz>
38452 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
38453 vector. Fix trailing white spaces.
38455 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
38457 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
38460 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38462 PR rtl-optimization/68664
38463 * config/arm/arm.c (arm_sched_can_speculate_insn):
38464 New function. Declare prototype.
38465 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38467 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38469 PR rtl-optimization/68664
38470 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
38472 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38474 2017-02-14 Amit Pawar <amit.pawar@amd.com>
38476 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
38477 max skip bytes for function, loop and jump.
38479 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38481 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
38482 ABS_EXPR for gimple dump.
38484 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38487 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
38489 PR tree-optimization/79408
38490 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
38491 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
38492 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
38493 also if rhs1 is INTEGER_CST.
38495 2017-02-14 Richard Biener <rguenther@suse.de>
38497 PR middle-end/79432
38498 * tree-into-ssa.c (insert_phi_nodes): When the function can
38499 have abnormal edges rewrite SSA names with broken use-def
38500 dominance out of SSA and register them for PHI insertion.
38502 2017-02-13 Martin Sebor <msebor@redhat.com>
38504 PR middle-end/79496
38505 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
38506 clearing info.nowrite flag when snprintf size argument is a range.
38508 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38510 * cprop.c (cprop_jump): Add missing space in string literal.
38511 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
38512 (get_constraint_for_component_ref): Likewise.
38513 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
38514 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
38515 * lra-constraints.c (process_alt_operands): Likewise.
38516 * ipa-inline.c (inline_small_functions): Likewise.
38517 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
38518 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
38519 * trans-mem.c (diagnose_tm_1_op): Likewise.
38520 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
38521 (grid_parallel_clauses_gridifiable): Likewise.
38523 * config/nvptx/mkoffload.c (process): Add space in between
38526 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
38527 "MOD4_SSE_REGS" and "ALL_REGS".
38529 * spellcheck.c (test_data): Add , in between "foo" and "food".
38531 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38534 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
38535 boundary crossing check and subsequent code generation agree.
38537 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38539 * config/aarch64/aarch64.c (has_memory_op): Delete.
38540 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
38543 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38545 PR rtl-optimization/79388
38546 PR rtl-optimization/79450
38547 * combine.c (distribute_notes): When removing TEM_INSN for which
38548 corresponding dest has last value recorded, invalidate that last
38551 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38553 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
38554 of explicit '@'. Add missing assembly comment marker on branch costs
38557 2017-02-13 Nathan Sidwell <nathan@acm.org>
38559 * gengtype-lex.l (<in_struct>): Add '/'.
38561 2017-02-13 Martin Liska <mliska@suse.cz>
38564 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
38566 2017-02-13 Richard Biener <rguenther@suse.de>
38568 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
38570 * configure: Re-generate.
38571 * config.in: Likewise.
38572 * graphite-dependences.c: Simplify as if
38573 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
38574 * graphite-isl-ast-to-gimple.c: Likewise.
38575 * graphite-optimize-isl.c: Likewise.
38576 * graphite-poly.c: Likewise.
38577 * graphite-sese-to-poly.c: Likewise.
38578 * graphite.h: Likewise.
38579 * toplev.c: Include isl/version.h and use isl_version () for
38580 printing the ISL version.
38581 * doc/install.texi: Update ISL requirement.
38583 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38585 * doc/standards.texi (Standards): Update reference to
38588 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38590 * doc/extend.texi (Named Address Spaces): sourceware.org now
38592 * doc/install.texi (Binaries): Ditto.
38595 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38597 * doc/cpp.texi: Replace "stringify"/"stringification" with C
38598 standard terminology "stringize"/"stringizing" throughout.
38599 * doc/cppinternals.texi: Likewise.
38601 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38603 * doc/extend.texi: Fix some spelling mistakes and typos.
38604 * doc/invoke.texi: Likewise.
38606 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38609 * params.def (inline-min-speedup) Change from 10 to 8.
38611 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38613 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
38616 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38619 * ipa-inline-analysis.c (get_minimal_bb): New function.
38620 (record_modified): Use it.
38621 (remap_edge_change_prob): Handle also ancestor functions.
38623 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
38625 * doc/contrib.texi (Contributors): Remove broken link into
38626 the Mauve CVS repository.
38628 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38630 PR middle-end/79454
38631 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
38632 result computation whenever lhs doesn't have vector mode, not
38633 just when it has BLKmode.
38635 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38637 * doc/makefile.texi (profiledbootstrap): Refer to the
38638 installation instructions only in textual form.
38640 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38643 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
38645 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38647 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
38648 (Specific): Update mingw-w64 reference.
38650 (Specific): Remove broken link to Renesas RX processor.
38652 2017-02-10 Richard Biener <rguenther@suse.de>
38654 * toplev.c (process_options): Do not mention obsolete graphite
38655 options when printing sorry message about missing graphite support.
38656 Mention -floop-nest-optimize.
38658 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
38660 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
38661 (vtst_p16): Likewise.
38662 (vtstq_p8): Likewise.
38663 (vtstq_p16): Likewise.
38665 (vtstq_p64): Likewise.
38666 * config/arm/arm_neon.h (vgetq_lane_p64): New.
38667 (vset_lane_p64): New.
38668 (vsetq_lane_p64): New.
38670 2017-02-10 Jakub Jelinek <jakub@redhat.com>
38672 PR tree-optimization/79411
38673 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
38674 stmt operands are SSA_NAMEs used in abnormal phis.
38675 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
38678 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38681 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
38684 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38686 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
38688 2017-02-09 Jakub Jelinek <jakub@redhat.com>
38690 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
38694 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
38695 not arbitrary TREE_CONSTANT.
38698 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
38699 "omp declare target link" attribute unless is_global_var.
38700 * omp-offload.c (find_link_var_op): Likewise.
38702 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
38703 Chung-Lin Tang <cltang@codesourcery.com>
38705 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
38707 (gimplify_adjust_omp_clauses): Don't delete TILE.
38708 (gimplify_omp_for): Deal with TILE.
38709 * internal-fn.c (expand_GOACC_TILE): New function.
38710 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
38712 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
38713 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
38715 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
38716 avoid DIV for outermost collapse var.
38717 (expand_oacc_for): Insert tile element loop as needed. Adjust.
38718 Remove out of date comments, fix whitespace.
38719 * omp-general.c (omp_extract_for_data): Deal with tiling.
38720 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
38721 adjust OLF_DIM_BASE value.
38722 (struct omp_for_data): Add tiling field.
38723 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
38724 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
38725 for auto loops. Remove default auto determining, moved to
38726 oacc_loop_fixed_partitions.
38727 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
38728 stmts, add e_mask field.
38729 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
38730 (oacc_thread_numbers): Use oacc_dim_call.
38731 (oacc_xform_tile): New.
38732 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
38733 (finish_oacc_loop): Adjust for ifns vector.
38734 (oacc_loop_discover_walk): Append loop abstraction sites to list,
38735 add case for GOACC_TILE fns.
38736 (oacc_loop_xform_loop): Delete.
38737 (oacc_loop_process): Iterate over call list directly, and add
38738 handling for GOACC_TILE fns.
38739 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
38741 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
38742 vector partitioning to outer loops. Assign 2 partitions to loops
38743 when available. Add TILE handling.
38744 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
38745 (execite_oacc_device_lower): Process GOACC_TILE fns,
38746 ignore unknown specs.
38747 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
38748 * tree.c (omp_clause_num_ops): Adjust TILE ops.
38749 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
38751 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
38753 * configure.ac (ACX_BUGURL): Update.
38754 * configure: Regenerate.
38756 2017-02-09 Richard Biener <rguenther@suse.de>
38758 PR tree-optimization/69823
38759 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
38760 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
38762 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38764 * config/arc/arc-c.def: Add __NPS400__ definition.
38765 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
38766 (TARGET_NPS400): Define.
38768 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38770 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
38772 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
38773 pointer, arch_info.
38774 (arc_cpu_types): Fill the arch_info field with a pointer into the
38775 arc_arch_types table.
38776 (arc_selected_cpu): Declare.
38777 * config/arc/arc.c (arc_selected_cpu): Make global.
38778 (arc_selected_arch): Delete.
38779 (arc_base_cpu): Delete.
38780 (arc_override_options): Remove references to deleted variables,
38781 update access to arch information.
38782 (ARC_OPT): Update access to arch information.
38783 (ARC_OPTX): Likewise.
38784 * config/arc/arc.h (arc_base_cpu): Remove declaration.
38785 (TARGET_ARC600): Update access to arch information.
38786 (TARGET_ARC601): Likewise.
38787 (TARGET_ARC700): Likewise.
38788 (TARGET_EM): Likewise.
38789 (TARGET_HS): Likewise.
38790 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
38793 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
38796 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
38797 condition/operands for integer GE/LE/GEU/LEU operations.
38799 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
38801 PR translation/79397
38802 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
38805 2017-02-08 Martin Jambor <mjambor@suse.cz>
38808 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38809 whether allocation happened.
38810 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38811 nothing was allocated.
38813 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38815 PR tree-optimization/79408
38816 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38817 constant, but SSA_NAME with a known integer range, use the minimum
38818 of that range instead of op1 to determine if modulo can be replaced
38819 with its first operand.
38821 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38823 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38825 2017-02-08 Richard Biener <rguenther@suse.de>
38827 PR tree-optimization/71824
38828 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38829 Check all loops contained in the merged region.
38831 2017-02-07 Andrew Pinski <apinski@cavium.com>
38833 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38835 2017-02-07 Andrew Pinski <apinski@cavium.com>
38837 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38838 (thunderxt88): Likewise.
38839 (thunderxt81): Disable LSE and change v8.1 to v8.
38840 (thunderxt83): Likewise.
38842 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38843 Richard Biener <rguenther@suse.de>
38845 PR middle-end/79399
38846 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38847 type from int to size_t.
38848 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38850 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38852 PR rtl-optimization/79386
38853 * cprop.c (bypass_conditional_jumps): Initialize
38854 bypass_last_basic_block already before splitting bbs after
38855 unconditional traps...
38856 (bypass_conditional_jumps): ... rather than here.
38859 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38860 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38861 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38862 fix -masm=intel patterns.
38864 2017-02-07 Richard Biener <rguenther@suse.de>
38866 PR tree-optimization/79256
38867 PR middle-end/79278
38868 * builtins.c (get_object_alignment_2): Use min_align_of_type
38869 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38870 and ADJUST_FIELD_ALIGN.
38872 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38874 * doc/tm.texi: Regenerate.
38875 * stor-layout.c (layout_decl): Adjust.
38876 (update_alignment_for_field): Likewise.
38877 (place_field): Likewise.
38878 (min_align_of_type): Likewise.
38879 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38880 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38881 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38882 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38883 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38884 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38885 * config/i386/i386.c (x86_field_alignment): Likewise.
38886 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38887 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38888 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38889 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38890 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38891 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38895 2017-01-30 Richard Biener <rguenther@suse.de>
38897 PR tree-optimization/79256
38898 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38899 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38902 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38904 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38905 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38906 builtins to SImode and emit a zero-extend, if necessary.
38908 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38910 * docs/invoke.texi (RISC-V Options): Alphabetize.
38912 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38914 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
38917 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38919 * config/riscv/riscv.c: New file.
38920 * common/config/riscv/riscv-common.c: Likewise.
38921 * config.gcc: Likewise.
38922 * config/riscv/constraints.md: Likewise.
38923 * config/riscv/elf.h: Likewise.
38924 * config/riscv/generic.md: Likewise.
38925 * config/riscv/linux.h: Likewise.
38926 * config/riscv/multilib-generator: Likewise.
38927 * config/riscv/peephole.md: Likewise.
38928 * config/riscv/pic.md: Likewise.
38929 * config/riscv/predicates.md: Likewise.
38930 * config/riscv/riscv-builtins.c: Likewise.
38931 * config/riscv/riscv-c.c: Likewise.
38932 * config/riscv/riscv-ftypes.def: Likewise.
38933 * config/riscv/riscv-modes.def: Likewise.
38934 * config/riscv/riscv-opts.h: Likewise.
38935 * config/riscv/riscv-protos.h: Likewise.
38936 * config/riscv/riscv.h: Likewise.
38937 * config/riscv/riscv.md: Likewise.
38938 * config/riscv/riscv.opt: Likewise.
38939 * config/riscv/sync.md: Likewise.
38940 * config/riscv/t-elf-multilib: Likewise.
38941 * config/riscv/t-linux: Likewise.
38942 * config/riscv/t-linux-multilib: Likewise.
38943 * config/riscv/t-riscv: Likewise.
38944 * configure.ac: Likewise.
38945 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
38946 Waterman as RISC-V maintainers.
38947 * doc/install.texi: Add RISC-V entries.
38948 * doc/invoke.texi: Add RISC-V options section.
38949 * doc/md.texi: Add RISC-V constraints section.
38950 * configure: Regenerated.
38952 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
38955 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
38956 false values to be constant vectors with all 0 or all 1 bits set.
38957 (vcondu<mode><mode>): Likewise.
38958 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
38960 (fpmask_comparison_operator): Update comment.
38961 (vecint_comparison_operator): New predicate.
38962 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
38963 vector conditionals when the true and false values are constant
38964 vectors with all 0 bits or all 1 bits set.
38966 2017-02-06 Martin Sebor <msebor@redhat.com>
38968 PR tree-optimization/79376
38969 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
38971 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
38973 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
38974 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
38975 to simplify split condition.
38977 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38979 * omp-expand.c (oxpand_omp_atomic_fetch_op,
38980 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
38983 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
38985 PR rtl-optimization/68664
38986 * target.def (can_speculate_insn): New hook.
38987 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
38988 * doc/tm.texi: Regenerate.
38989 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
38990 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
38991 (rs6000_sched_can_speculate_insn): New function.
38993 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38995 PR tree-optimization/79284
38996 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
38997 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
38998 vectorizable_mask_load_store, vectorizable_operation,
38999 vect_is_simple_cond, get_same_sized_vectype): Use it instead
39000 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
39001 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
39002 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
39003 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
39004 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
39005 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
39006 is_gimple_assign (stmt). Replace another such test with
39007 is_gimple_assign (stmt).
39009 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
39012 * config/avr/avr.c (rtl-iter.h): Include it.
39013 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
39014 (avr_legitimate_combined_insn): ...and implementation.
39016 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
39018 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
39019 * config/s390/s390.c (s390_const_operand_ok)
39020 (s390_canonicalize_comparison, s390_extract_part)
39021 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
39022 (s390_contiguous_bitmask_p, s390_rtx_costs)
39023 (legitimize_pic_address): Likewise.
39024 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
39025 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
39026 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
39027 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
39028 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
39030 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
39032 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
39033 REGNO($0) == REGNO($1).
39035 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39037 * config/s390/linux.h(SIZE_TYPE): Add comment.
39039 2017-02-06 Julian Brown <julian@codesourcery.com>
39040 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
39041 Virendra Pathak <virendra.pathak@broadcom.com>
39043 * config/aarch64/aarch64-cores.def: Change the scheduler
39045 * config/aarch64/aarch64.md: Include thunderx2t99.md.
39046 * config/aarch64/thunderx2t99.md: New file.
39048 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39050 * doc/standards.texi (Go Language): Update link to language
39053 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
39055 * tree-eh.c (lower_resx): Sanitize profile.
39056 (cleanup_empty_eh_move_lp): Likewise.
39058 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
39061 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
39063 * cfgloopmanip.h (loop_version): Update prototype.
39064 * modulo-sched.c (sms_schedule): Update call of loop_version.
39065 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
39066 * tree-parloops.c (gen_parallel_loop): Likewise.
39067 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
39068 * tree-ssa-loop-split.c (split_loop): Likewise.
39069 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
39070 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
39072 2017-02-05 Martin Liska <mliska@suse.cz>
39075 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
39077 (print_operand_address): Initialize a struct to zero.
39079 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39081 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
39082 garbage collector only in textual form.
39084 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39086 * doc/extend.texi (x86 specific memory model extensions for
39087 transactional memory): Simplify a phrase.
39089 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
39092 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
39093 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
39094 (atomic_storedi_1): Likewise.
39096 2017-02-04 Jakub Jelinek <jakub@redhat.com>
39098 PR tree-optimization/79338
39099 * tree-parloops.c (gather_scalar_reductions): Don't call
39100 vect_analyze_loop_form for loop->inner before destroying loop's
39103 2017-02-03 Martin Sebor <msebor@redhat.com>
39105 PR tree-optimization/79327
39106 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
39107 when precision has resulted in leading zeros.
39108 (format_integer): Adjust the likely counter to assume an unknown
39109 argument that may be zero is non-zero.
39111 2017-02-03 Jason Merrill <jason@redhat.com>
39114 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
39115 avoid copying non-taken branch.
39117 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39119 PR tree-optimization/79340
39120 * tree-vect-loop.c (vectorizable_reduction): Release
39121 vec_defs elements after safe_splicing them into other vectors.
39124 PR tree-optimization/79327
39125 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
39126 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
39128 (format_integer): Use wide_int_to_tree instead of build_int_cst
39129 + to_?hwi. If argmin is NULL, just set argmin and argmax to
39130 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
39131 of shortest and longest sequence.
39133 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
39135 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
39136 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
39138 2017-02-03 Walter Lee <walt@tilera.com>
39141 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
39142 after initial stackframe link reg save.
39143 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
39145 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39148 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
39149 wu for stxssp alternative.
39151 2017-02-03 Martin Sebor <msebor@redhat.com>
39153 PR tree-optimization/79352
39154 * gimple-fold.c (get_range_strlen): Add argument.
39155 (get_range_strlen): Change return type to bool.
39156 (get_maxval_strlen): Pass in a dummy argument.
39157 * gimple-fold.h (get_range_strlen): Change return type to bool.
39158 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
39159 * tree.h (array_at_struct_end_p): Add argument.
39160 * tree.c (array_at_struct_end_p): Handle it.
39162 2017-02-03 Martin Liska <mliska@suse.cz>
39165 * multiple_target.c (create_dispatcher_calls): Redirect edge
39166 from a caller of a dispatcher.
39167 (expand_target_clones): Make the clones local.
39168 (ipa_target_clone): Do both target clones and resolvers.
39169 (ipa_dispatcher_calls): Remove the pass.
39170 (pass_dispatcher_calls::gate): Likewise.
39171 (make_pass_dispatcher_calls): Likewise.
39172 * passes.def (pass_target_clone): Put as very first IPA early
39175 2017-02-03 Martin Liska <mliska@suse.cz>
39177 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
39178 in case of a function with ifunc attribute.
39180 2017-02-03 Martin Liska <mliska@suse.cz>
39182 * cgraph.c (cgraph_node::dump): Dump function version info.
39183 * symtab.c (symtab_node::dump_base): Add missing new line.
39185 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39187 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
39188 (ifcombine_ifandif): Use it.
39190 2017-02-03 Martin Liska <mliska@suse.cz>
39192 * doc/invoke.texi: Document default value for
39193 use-after-scope-direct-emission-threshold.
39195 2017-02-03 Martin Liska <mliska@suse.cz>
39197 PR tree-optimization/79339
39198 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
39199 (format_floating): Likewise.
39201 2017-02-03 Martin Liska <mliska@suse.cz>
39204 * ipa-prop.c (ipa_node_params_t::insert): Remove current
39206 (ipa_node_params_t::remove): Likewise.
39207 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
39208 initialization from removed ipa_node_params_t::insert.
39209 (ipa_node_params::~ipa_node_params): Move from removed
39210 ipa_node_params_t::release.
39211 * symbol-summary.h (symbol_summary::m_released): New member.
39212 Do not release a summary twice. Do not allow to call finalizer
39213 for types of a summary that live in GGC memory.
39215 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
39217 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
39220 2017-02-02 Martin Sebor <msebor@redhat.com>
39222 PR middle-end/79275
39223 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
39224 (format_string): Tighten up the range of output for non-constant
39225 strings and correct the expected range for wide non-constant strings.
39227 2017-02-02 Martin Sebor <msebor@redhat.com>
39229 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
39231 PR middle-end/32003
39232 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
39234 (-fdump-tree-@var): Add to index and document how to come up
39235 with pass-specific option and dump file names.
39236 (-fdump-passes): Clarify where to look for output.
39238 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39240 PR middle-end/77445
39241 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
39242 statistics of the analyzed path; allow threading for speed when
39243 any of BBs along the path are optimized for speed.
39245 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
39247 PR middle-end/78468
39248 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
39249 settings of the virtual registers.
39252 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
39254 * explow.c (get_dynamic_stack_size): Take known alignment of stack
39255 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
39258 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39260 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
39261 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
39263 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39265 * config/s390/s390.md: Add missing comments with the expanded
39267 * config/s390/vector.md: Likewise.
39268 * config/s390/vx-builtins.md: Likewise.
39270 2017-02-02 Jakub Jelinek <jakub@redhat.com>
39273 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
39274 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
39275 conditions on a single line.
39277 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39279 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39280 __S390_VX__ to __VX__.
39282 2017-02-01 Andrew Pinski <apinski@cavium.com>
39284 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
39285 stmt_info to record_stmt_cost.
39286 (vect_get_known_peeling_cost): Pass stmt_info if known to
39288 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
39289 cpu_vector_cost field into
39290 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
39291 field into vec_int_stmt_cost and vec_fp_stmt_cost.
39292 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
39293 splitting of scalar_stmt_cost and vec_stmt_cost.
39294 (thunderx_vector_cost): Likewise.
39295 (cortexa57_vector_cost): LIkewise.
39296 (exynosm1_vector_cost): Likewise.
39297 (xgene1_vector_cost): Likewise.
39298 (thunderx2t99_vector_cost): Improve after the splitting of the two
39300 (aarch64_builtin_vectorization_cost): Update for the splitting of
39301 scalar_stmt_cost and vec_stmt_cost.
39303 2017-02-01 Torvald Riegel <triegel@redhat.com>
39304 Richard Henderson <rth@redhat.com>
39306 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
39307 conditional on existance of a fast atomic load.
39308 * optabs-query.c (can_atomic_load_p): New function.
39309 * optabs-query.h (can_atomic_load_p): Declare it.
39310 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
39311 no fast atomic load is available for the particular size of access.
39312 (expand_atomic_compare_and_swap): Likewise.
39313 (expand_atomic_load): Likewise.
39314 (expand_atomic_store): Likewise.
39315 (expand_atomic_fetch_op): Likewise.
39316 * testsuite/lib/target-supports.exp
39317 (check_effective_target_sync_int_128): Remove x86 because it provides
39318 no fast atomic load.
39319 (check_effective_target_sync_int_128_runtime): Likewise.
39321 2017-02-01 Richard Biener <rguenther@suse.de>
39323 * graphite.c: Include tree-vectorizer.h for find_loop_location.
39324 (graphite_transform_loops): Provide opt-info for optimized nests.
39325 * tree-parloop.c (parallelize_loops): Provide opt-info for
39326 parallelized loops.
39328 2017-02-01 Richard Biener <rguenther@suse.de>
39330 PR middle-end/79315
39331 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
39332 was not set before.
39334 2017-02-01 Richard Biener <rguenther@suse.de>
39336 PR tree-optimization/71824
39337 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
39338 Verify the loops are valid in the merged SESE region.
39339 (scop_detection::can_represent_loop_1): Check analyzing the
39340 evolution of the number of iterations in the region succeeds.
39342 2017-01-31 Ian Lance Taylor <iant@golang.org>
39344 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
39345 REG_ARGS_SIZE note to 32-bit push insns and call insn.
39347 2017-01-31 David Malcolm <dmalcolm@redhat.com>
39349 PR preprocessor/79210
39350 * input.c (get_substring_ranges_for_loc): Replace line_width
39351 assertion with error-handling.
39353 2017-01-31 Richard Biener <rguenther@suse.de>
39355 PR tree-optimization/77318
39356 * graphite-sese-to-poly.c (extract_affine): Fix assert.
39357 (create_pw_aff_from_tree): Take loop parameter.
39358 (add_condition_to_pbb): Pass loop of the condition to
39359 create_pw_aff_from_tree.
39361 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39363 * config/s390/s390.c (s390_asan_shadow_offset): New function.
39364 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
39366 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
39370 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
39372 (convert_int_to_float128): Likewise.
39373 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
39374 (convert_int_to_float128): Likewise.
39375 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
39376 (UNSPEC_IEEE128_CONVERT): Likewise.
39377 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
39378 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
39379 Use local variables for IBM extended format.
39380 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
39381 (fix_trunc<mode>si2_fprs): Likewise.
39382 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
39383 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
39384 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
39385 to know that we can now have integers of all sizes in vector
39387 (fix<uns>_<mode>di2_hw): Likewise.
39388 (float<uns>_<mode>si2_hw): Likewise.
39389 (fix_<mode>si2_hw): Likewise.
39390 (fixuns_<mode>si2_hw): Likewise.
39391 (float<uns>_<mode>di2_hw): Likewise.
39392 (float_<mode>di2_hw): Likewise.
39393 (float_<mode>si2_hw): Likewise.
39394 (floatuns_<mode>di2_hw): Likewise.
39395 (floatuns_<mode>si2_hw): Likewise.
39396 (xscvqp<su>wz_<mode>): Delete, no longer used.
39397 (xscvqp<su>dz_<mode>): Likewise.
39398 (xscv<su>dqp_<mode>): Likewise.
39399 (ieee128_mfvsrd_64bit): Likewise.
39400 (ieee128_mfvsrd_32bit): Likewise.
39401 (ieee128_mfvsrwz): Likewise.
39402 (ieee128_mtvsrw): Likewise.
39403 (ieee128_mtvsrd_64bit): Likewise.
39404 (ieee128_mtvsrd_32bit): Likewise.
39406 2017-01-31 Martin Liska <mliska@suse.cz>
39409 * ipa-prop.c (ipa_free_all_node_params): Call release method
39410 instead of ~sumbol_summary to not to trigger double times
39413 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
39415 PR tree-optimization/71691
39416 * bitmap.h (class auto_bitmap): New.
39417 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
39418 is_maybe_undefined instead of ssa_undefined_value_p.
39420 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39422 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39423 __S390_ARCH_LEVEL__ to __ARCH__.
39425 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39427 PR tree-optimization/79267
39428 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
39429 if should_remove_lhs_p is true.
39431 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
39434 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
39435 (add_alignment_attribute): New.
39436 (base_type_die): Add alignment attribute.
39437 (subrange_type_die): Likewise.
39438 (modified_type_die): Likewise.
39439 (gen_array_type_die): Likewise.
39440 (gen_descr_array_type_die: Likewise.
39441 (gen_enumeration_type_die): Likewise.
39442 (gen_subprogram_die): Likewise.
39443 (gen_variable_die): Likewise.
39444 (gen_field_die): Likewise.
39445 (gen_ptr_to_mbr_type_die): Likewise.
39446 (gen_struct_or_union_type_die): Likewise.
39447 (gen_subroutine_type_die): Likewise.
39448 (gen_typedef_die): Likewise.
39449 (base_type_cmp): Compare alignment attribute.
39451 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39454 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
39455 (setb_unsigned) New pattern for setb with CCUNS.
39456 * config/rs6000/rs6000.c (expand_block_compare): Use a different
39457 subfc./subfe sequence to avoid overflow problems. Generate a
39458 shorter sequence with cmpld/setb for power9.
39459 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
39460 for generating subfc. instruction.
39461 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
39462 now uses this instruction.
39464 2017-01-30 Ian Lance Taylor <iant@google.com>
39467 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
39468 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
39470 2017-01-30 Martin Sebor <msebor@redhat.com>
39472 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
39473 Move constant to the right of a relational operator.
39474 (get_mpfr_format_length, format_character, format_string): Ditto.
39475 (should_warn_p, maybe_warn): Same.
39477 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
39479 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
39482 * asan.c (get_translation_unit_decl): Remove function.
39483 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
39485 2017-01-30 Martin Liska <mliska@suse.cz>
39487 PR gcov-profile/79259
39488 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
39489 -fprofile-generate.
39491 2017-01-30 Martin Liska <mliska@suse.cz>
39494 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
39495 Initialize variables with NULL value.
39497 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
39500 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
39502 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
39504 2017-01-30 Richard Biener <rguenther@suse.de>
39506 PR tree-optimization/79276
39507 * tree-vrp.c (process_assert_insertions): Properly adjust common
39508 when removing a duplicate.
39510 2017-01-30 Richard Biener <rguenther@suse.de>
39512 PR tree-optimization/79256
39513 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
39514 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
39516 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
39518 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
39521 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
39522 ("*r<noxa>sbg_<mode>_sll_bitmask")
39523 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
39524 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
39525 Use contiguous_bitmask_nowrap_operand.
39527 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39530 * config/rs6000/altivec.h (vec_xl): Revise #define.
39531 (vec_xst): Likewise.
39533 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
39535 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
39537 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
39539 PR rtl-optimization/79194
39540 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
39541 traps before call to bypass_conditional_jumps.
39543 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39545 PR tree-optimization/71374
39546 * lra-constraints.c (check_conflict_input_operands): New.
39547 (match_reload): Use it.
39549 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39552 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
39553 account to calculate conflict_set.
39555 2017-01-27 Bin Cheng <bin.cheng@arm.com>
39557 PR rtl-optimization/78559
39558 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
39559 other_insn in combine.
39561 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
39563 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
39564 uint16_type_node for BT_UINT16.
39566 2017-01-27 David Malcolm <dmalcolm@redhat.com>
39568 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
39569 "RTL Tests" to menu.
39570 (GIMPLE Tests): New node.
39571 (RTL Tests): New node.
39573 2017-01-27 Richard Biener <rguenther@suse.de>
39575 PR tree-optimization/79245
39576 * tree-loop-distribution.c (distribute_loop): Apply cost
39577 modeling also to detected patterns.
39579 2017-01-27 Richard Biener <rguenther@suse.de>
39581 PR tree-optimization/71433
39582 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
39583 (compare_assert_loc): New function.
39584 (process_assert_insertions): Sort and optimize assert locations
39585 to remove duplicates and push down identical assertions on
39586 edges to their destination block.
39588 2017-01-27 Richard Biener <rguenther@suse.de>
39590 PR tree-optimization/79244
39591 * tree-vrp.c (remove_range_assertions): Forcefully propagate
39592 out SSA names even if abnormal.
39594 2017-01-27 Jakub Jelinek <jakub@redhat.com>
39596 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
39597 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
39598 instead of MPFR_RNDN.
39600 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
39603 * arm.c (arm_option_override): Don't call build_target_option_node
39604 until after doing all option overrides.
39605 (arm_valid_target_attribute_tree): Likewise.
39607 2017-01-27 Martin Liska <mliska@suse.cz>
39609 * doc/invoke.texi (-fprofile-arcs): Document profiling support
39610 for {cd}tors and C++ {cd}tors.
39612 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
39614 * config/s390/s390.md ("*setmem_long_and")
39615 ("*setmem_long_and_31z"): Use zero_extend instead of and.
39617 2017-01-26 Martin Sebor <msebor@redhat.com>
39619 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
39622 2017-01-26 Martin Sebor <msebor@redhat.com>
39624 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
39625 HAVE_DFmode before using XFmode or DFmode.
39626 (parse_directive): Avoid using the z length modifier to avoid
39627 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
39629 PR middle-end/78703
39630 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
39631 to accept adjustment as an array.
39632 (get_int_range): New function.
39633 (struct directive): Make width and prec arrays.
39634 (directive::set_width, directive::set_precision): Call get_int_range.
39635 (format_integer, format_floating): Handle width and precision ranges.
39636 (format_string, parse_directive): Same.
39638 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39641 * dwarf2out.c (generate_skeleton_bottom_up): For children with
39642 comdat_type_p set, just clone them, but keep the children in the
39646 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
39647 which have direct callers with -fvar-tracking-assignments enabled
39649 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
39650 inside of type units.
39652 2017-01-26 Martin Sebor <msebor@redhat.com>
39654 PR middle-end/78703
39655 * gimple-ssa-sprintf.c (struct result_range): Add likely and
39657 (struct format_result): Replace number_chars, number_chars_min,
39658 and number_chars_max with a single member of struct result_range.
39660 (format_result::operator+=): Adjust.
39661 (struct fmtresult): Remove bounded. Handle likely and unlikely
39663 (fmtresult::adjust_for_width_or_precision): New function.
39664 (fmtresult:type_max_digits): New function.
39665 (bytes_remaining): Handle likely and unlikely counters.
39666 (min_bytes_remaining): Remove.
39667 (format_percent): Simplify.
39668 (format_integer, format_floating): Set likely and unlikely counters.
39669 (get_string_length, format_character, format_string): Same.
39670 (format_plain, should_warn_p): New function.
39671 (maybe_warn): Call should_warn_p. Update diagnostic messages
39672 and handle those for all directives, including plain strings.
39673 (format_directive): Handle likely and unlikely counters.
39674 Remove unnecessary quoting from diagnostics. Add an informational
39676 (add_bytes): Remove.
39677 (pass_sprintf_length::compute_format_length): Simplify.
39678 (try_substitute_return_value): Handle likely and unlikely counters.
39680 2017-01-26 Carl Love <cel@us.ibm.com>
39682 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
39683 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
39685 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
39688 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
39689 endianess for subregs into account.
39690 * lra-constraints.c (lra_constraints): Do risky transformations
39691 always on the first iteration.
39692 * lra-lives.c (check_pseudos_live_through_calls): Add arg
39693 last_call_used_reg_set.
39694 (process_bb_lives): Define and use last_call_used_reg_set.
39695 * lra.c (lra): Always continue after lra_constraints on the first
39698 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
39700 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
39702 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
39704 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39706 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
39707 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
39708 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
39709 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
39710 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
39711 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
39712 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
39713 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
39714 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
39716 2017-01-26 Marek Polacek <polacek@redhat.com>
39719 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
39720 for the third operand.
39722 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39724 PR middle-end/79236
39725 * omp-low.c (struct omp_context): Add simt_stmt field.
39726 (scan_omp_for): Return omp_context *.
39727 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
39728 context to the _simt_ SIMD stmt.
39729 (lower_omp_for): For combined SIMD with sibling _simt_
39730 SIMD, make sure to use the same decls in _looptemp_
39731 clauses as in the sibling.
39733 2017-01-26 David Sherwood <david.sherwood@arm.com>
39735 PR middle-end/79212
39736 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
39739 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39742 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
39743 emit fld b; fld a; if possible.
39745 * brig-builtins.def: Update copyright years.
39746 * config/arm/arm_acle_builtins.def: Update copyright years.
39748 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
39751 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
39752 constraint instead of o for the stxsd instruction.
39754 2017-01-25 Carl Love <cel@us.ibm.com>
39756 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
39757 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
39759 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
39761 * doc/invoke.texi (C++ Dialect Options): Fix typo.
39763 2017-01-25 Richard Biener <rguenther@suse.de>
39765 PR tree-optimization/69264
39766 * target.def (vector_alignment_reachable): Improve documentation.
39767 * doc/tm.texi: Regenerate.
39768 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
39770 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
39771 earlier changes with respect to TYPE_USER_ALIGN.
39772 (vector_alignment_reachable_p): Likewise. Improve dumping.
39774 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39777 * config/arm/arm.md (xordi3): Force constant operand into a register
39780 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39782 * doc/invoke.texi (-fstore-merging): Correct default optimization
39783 levels at which it is enabled.
39784 (-O): Move -fstore-merging from list to...
39787 2017-01-25 Richard Biener <rguenther@suse.de>
39790 * omp-expand.c: Include debug.h.
39791 (expand_omp_taskreg): Make sure to generate early debug before
39792 outlining anything from a function.
39793 (expand_omp_target): Likewise.
39794 (grid_expand_target_grid_body): Likewise.
39796 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
39799 * asan.c (get_translation_unit_decl): New function.
39800 (asan_add_global): Extract modules file name from globals
39801 TRANSLATION_UNIT_DECL name.
39803 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39806 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39807 for long calls with APCS frame and VFP.
39809 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39811 * cfg.c (original_copy_tables_initialized_p): New function.
39812 * cfg.h (original_copy_tables_initialized_p): New decl.
39813 * cfgrtl.c (relink_block_chain): Guard the call to
39814 free_original_copy_tables with a call to
39815 original_copy_tables_initialized_p.
39816 * cgraph.h (symtab_node::native_rtl_p): New decl.
39817 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39818 (symtab_node::needed_p): Don't assert for early assembly output
39819 for __RTL functions.
39820 (cgraph_node::finalize_function): Set "force_output" for __RTL
39822 (cgraph_node::analyze): Bail out early for __RTL functions.
39823 (analyze_functions): Update assertion to support __RTL functions.
39824 (cgraph_node::expand): Bail out early for __RTL functions.
39825 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39827 * function.h (struct function): Update comment for field
39829 * gimple-expr.c: Include "tree-pass.h".
39830 (gimple_has_body_p): Return false for __RTL functions.
39831 * Makefile.in (OBJS): Add run-rtl-passes.o.
39832 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39834 (gcc::pass_manager::get_clean_slate): New accessor.
39835 * passes.c: Include "insn-addr.h".
39836 (should_skip_pass_p): Add logging. Update logic for running
39837 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39838 property-provider override so it is only done for gimple passes.
39840 (skip_pass): New function.
39841 (execute_one_pass): Call skip_pass when skipping passes.
39842 * read-md.c (md_reader::read_char): Support filtering
39843 the input to a subset of line numbers.
39844 (md_reader::md_reader): Initialize fields
39845 m_first_line and m_last_line.
39846 (md_reader::read_file_fragment): New function.
39847 * read-md.h (md_reader::read_file_fragment): New decl.
39848 (md_reader::m_first_line): New field.
39849 (md_reader::m_last_line): New field.
39850 * read-rtl-function.c (function_reader::create_function): Only
39851 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39852 curr_properties. Set DECL_INITIAL to a dummy block.
39853 (read_rtl_function_body_from_file_range): New function.
39854 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39856 * run-rtl-passes.c: New file.
39857 * run-rtl-passes.h: New file.
39859 2017-01-24 Jeff Law <law@redhat.com>
39861 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39864 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39866 PR tree-optimization/79159
39867 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39868 (record_nonwrapping_iv): Improve boundary using above function if no
39869 value range information.
39871 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39872 Martin Jambor <mjambor@suse.cz>
39874 * brig-builtins.def: New file.
39875 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39876 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39877 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39878 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39879 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39880 * builtin-types.def (BT_INT8): New.
39881 (BT_INT16): Likewise.
39882 (BT_UINT8): Likewise.
39883 (BT_UINT16): Likewise.
39884 (BT_FN_ULONG): Likewise.
39885 (BT_FN_UINT_INT): Likewise.
39886 (BT_FN_UINT_ULONG): Likewise.
39887 (BT_FN_UINT_LONG): Likewise.
39888 (BT_FN_UINT_PTR): Likewise.
39889 (BT_FN_ULONG_PTR): Likewise.
39890 (BT_FN_INT8_FLOAT): Likewise.
39891 (BT_FN_INT16_FLOAT): Likewise.
39892 (BT_FN_UINT32_FLOAT): Likewise.
39893 (BT_FN_UINT16_FLOAT): Likewise.
39894 (BT_FN_UINT8_FLOAT): Likewise.
39895 (BT_FN_UINT64_FLOAT): Likewise.
39896 (BT_FN_UINT16_UINT32): Likewise.
39897 (BT_FN_UINT32_UINT16): Likewise.
39898 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39899 (BT_FN_INT_PTR_INT): Likewise.
39900 (BT_FN_UINT_PTR_UINT): Likewise.
39901 (BT_FN_LONG_PTR_LONG): Likewise.
39902 (BT_FN_ULONG_PTR_ULONG): Likewise.
39903 (BT_FN_VOID_UINT64_UINT64): Likewise.
39904 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39905 (BT_FN_INT8_INT8_INT8): Likewise.
39906 (BT_FN_INT16_INT16_INT16): Likewise.
39907 (BT_FN_INT_INT_INT): Likewise.
39908 (BT_FN_UINT_FLOAT_UINT): Likewise.
39909 (BT_FN_FLOAT_UINT_UINT): Likewise.
39910 (BT_FN_ULONG_UINT_UINT): Likewise.
39911 (BT_FN_ULONG_UINT_PTR): Likewise.
39912 (BT_FN_ULONG_ULONG_ULONG): Likewise.
39913 (BT_FN_UINT_UINT_UINT): Likewise.
39914 (BT_FN_VOID_UINT_PTR): Likewise.
39915 (BT_FN_UINT_UINT_PTR: Likewise.
39916 (BT_FN_UINT32_UINT64_PTR): Likewise.
39917 (BT_FN_INT_INT_UINT_UINT): Likewise.
39918 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
39919 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
39920 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
39921 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
39922 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
39923 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
39924 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
39925 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
39926 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
39927 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
39928 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
39929 * doc/frontends.texi: List BRIG FE.
39930 * doc/install.texi (Testing): Add BRIG tesring requirements.
39931 * doc/invoke.texi (Overall Options): Mention BRIG.
39932 * doc/standards.texi (Standards): Doucment BRIG HSA version.
39934 2017-01-24 Richard Biener <rguenther@suse.de>
39936 PR translation/79208
39937 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
39939 2017-01-24 Martin Jambor <mjambor@suse.cz>
39942 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
39943 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
39944 and known_contexts.
39946 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
39948 PR middle-end/79123
39949 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
39950 casts from signed to unsigned really don't have a range.
39952 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
39954 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
39955 GMP_RNDx for compatiblity.
39957 2017-01-24 Martin Liska <mliska@suse.cz>
39960 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
39961 that would prevent us to call alloca with -1 as argument.
39963 2017-01-24 Jakub Jelinek <jakub@redhat.com>
39965 * dwarf2out.c (output_compilation_unit_header, output_file_names):
39966 Avoid -Wformat-security warning.
39968 2017-01-23 Andrew Pinski <apinski@cavium.com>
39970 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
39973 2017-01-23 Martin Sebor <msebor@redhat.com>
39975 PR middle-end/78703
39976 * gimple-ssa-sprintf.c (warn_level): New global.
39977 (format_integer): Use it here and throughout the rest of the file.
39978 Use the same switch to compute sign as base.
39979 (maybe_warn): New function.
39980 (format_directive): Factor out warnings into maybe_warn.
39981 Add debugging output. Use warn_level.
39982 (add_bytes): Use warn_level.
39983 (pass_sprintf_length::compute_format_length): Add debugging output.
39984 (try_substitute_return_value): Same.
39985 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
39987 PR middle-end/78703
39988 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
39989 (struct fmtresult, format_integer, format_floating): Adjust.
39990 (fmtresult::fmtresult): Set max correctly in two argument ctor.
39991 (get_string_length, format_string,format_directive): Same.
39992 (pass_sprintf_length::compute_format_length): Same.
39993 (try_substitute_return_value): Simplify slightly.
39995 PR middle-end/78703
39996 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
39997 (fmtresult::operator+=): Outlined.
39998 (struct fmtresult): Add ctors.
39999 (struct conversion_spec): Rename...
40000 (struct directive): ...to this. Add and remove data members.
40001 (directive::set_width, directive::set_precision): New functions.
40002 (format_percent): Use fmtresult ctor.
40003 (get_width_and_precision): Remove.
40004 (format_integer): Make naming changes. Avoid computing width and
40006 (format_floating): Same. Adjust indentation.
40007 (format_character, format_none): New functions.
40008 (format_string): Moved character handling to format_character.
40009 (format_directive): Remove arguments, change return type.
40010 (parse_directive): New function.
40011 (pass_sprintf_length::compute_format_length): Move directive
40012 parsing to parse_directive.
40014 2017-01-23 Jakub Jelinek <jakub@redhat.com>
40016 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
40017 (assign_assembler_name_if_needed): ... this.
40018 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
40019 (assign_assembler_name_if_needed): ... this.
40020 (free_lang_data_in_cgraph): Adjust callers.
40021 * cgraphunit.c (cgraph_node::analyze): Likewise.
40022 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
40025 2017-01-23 Richard Biener <rguenther@suse.de>
40027 PR tree-optimization/79088
40028 PR tree-optimization/79188
40029 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
40030 resetting loop bounds after last path deletion. Reset loop
40031 bounds of the target loop, make code match the comments.
40032 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
40033 Make sure loops need no fixups.
40035 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
40037 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
40038 exponent support with double type for first argument.
40039 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
40040 type returned by __builtin_vec_extract_sig,
40041 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
40042 functions from "vector int" to "vector unsigned int" or from
40043 "vector long long int" to "vector unsigned long long int".
40044 Changed type returned by __builtin_vec_extract_exp,
40045 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
40046 functions from "vector int" to "vector unsigned int" or from
40047 "vector long long int" to "vector unsigned long long int".
40048 Changed return type of __builtin_vec_test_data_class,
40049 __builtin_vec_test_data_class_sp, and
40050 __builtin_vec_test_data_class_dp from "vector int" to
40051 "vector bool int" or from "vector long long int" to "vector bool
40052 long long int" and changed second argument type from "unsigned
40053 int" to "int". Added new overloaded function forms "vector float
40054 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
40055 "vector float __builtin_vec_insert_exp_sp (vector float, vector
40056 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
40057 double, vector unsigned long long int)" and "vector double
40058 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
40059 long int)". Changed return type of
40060 __builtin_scalar_test_data_class and
40061 __builtin_scalar_test_data_class_sp and
40062 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
40063 int" and changed second argument from "unsigned int" to "int".
40064 Changed type returned by __builtin_scalar_test_neg,
40065 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
40066 from "int" to "bool int". Added new overloaded function form
40067 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
40068 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
40069 exponent double-precision with floating point first argument.
40070 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
40071 documentation of scalar_test_data_class, scalar_test_neg,
40072 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
40073 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
40074 vec_test_data_class built-in functions to reflect refinements in
40075 their type signatures.
40077 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
40079 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
40081 (aarch64_elf_asm_destructor): Likewise.
40083 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
40085 PR rtl-optimization/78634
40086 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
40087 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
40088 * ifcvt.c (noce_try_cmove): Add missing cost check.
40090 PR rtl-optimization/71724
40091 * combine.c (if_then_else_cond): Look for situations where it is
40092 beneficial to undo the work of one of the recursive calls.
40094 2017-01-23 Bin Cheng <bin.cheng@arm.com>
40096 PR tree-optimization/70754
40097 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
40098 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
40099 combined stmt before it if not NULL.
40100 (combine_chains): Process refs reversely and compute dominance point
40103 2017-01-23 Martin Liska <mliska@suse.cz>
40105 PR tree-optimization/79196
40106 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
40107 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
40109 (strlen_optimize_stmt): Call the renamed function.
40111 2017-01-23 Michael Matz <matz@suse.de>
40113 PR tree-optimization/78384
40114 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
40116 2017-01-23 Richard Biener <rguenther@suse.de>
40118 PR tree-optimization/79186
40119 * tree-vrp.c (register_new_assert_for): Make sure we've seen
40120 both incoming edges before moving an assert.
40122 2017-01-23 Martin Jambor <mjambor@suse.cz>
40124 * ipa-prop.c (load_from_param_1): Removed.
40125 (load_from_unmodified_param): Bits from load_from_param_1 put back
40127 (load_from_param): Removed.
40128 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
40129 with stmt. Reverted back to use of load_from_unmodified_param.
40131 2017-01-23 Martin Jambor <mjambor@suse.cz>
40134 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
40135 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
40136 field a pointer to garbage collected vector, mark lattices and
40137 ipcp_orig_node with GTY((skip)).
40138 (ipa_get_param_count): Adjust to descriptors being a pointer.
40139 (ipa_get_param): Likewise.
40140 (ipa_get_type): Likewise.
40141 (ipa_get_param_move_cost): Likewise.
40142 (ipa_set_param_used): Likewise.
40143 (ipa_get_controlled_uses): Likewise.
40144 (ipa_set_controlled_uses): Likewise.
40145 (ipa_is_param_used): Likewise.
40146 (ipa_node_params_t): Move into garbage collector. New methods insert
40148 (ipa_node_params_sum): Annotate wth GTY(()).
40149 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
40151 (ipa_load_from_parm_agg): Adjust declaration.
40152 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
40153 * ipa-profile.c (ipa_profile): Likewise.
40154 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
40155 (ipa_populate_param_decls): Make descriptors parameter garbage
40157 (ipa_dump_param): Adjust to descriptors being a pointer.
40158 (ipa_alloc_node_params): Likewise.
40159 (ipa_initialize_node_params): Likewise.
40160 (load_from_param_1): Make descriptors parameter garbage collected.
40161 (load_from_unmodified_param): Likewise.
40162 (load_from_param): Likewise.
40163 (ipa_load_from_parm_agg): Likewise.
40164 (ipa_node_params::~ipa_node_params): Removed.
40165 (ipa_free_all_node_params): Remove call to delete operator.
40166 (ipa_node_params_t::insert): New.
40167 (ipa_node_params_t::remove): Likewise.
40168 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
40169 copy known_csts and known_contexts vectors.
40170 (ipa_read_node_info): Adjust to descriptors being a pointer.
40171 (ipcp_modif_dom_walker): Make m_descriptors field garbage
40173 (ipcp_transform_function): Make descriptors variable garbage
40176 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
40178 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
40179 * config/i386/avx512dqintrin.h: Ditto.
40180 * config/i386/avx512fintrin.h: Ditto.
40181 * config/i386/i386.c: Handle new builtins.
40182 * config/i386/i386-builtin.def: Add new builtins.
40183 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
40184 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
40186 2017-01-23 Jakub Jelinek <jakub@redhat.com>
40187 Martin Liska <mliska@suse.cz>
40189 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
40190 * asan.c (asan_expand_poison_ifn): Support stores and use
40191 appropriate ASAN report function.
40192 * internal-fn.c (expand_ASAN_POISON_USE): New function.
40193 * internal-fn.def (ASAN_POISON_USE): Declare.
40194 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
40195 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
40196 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
40197 ASAN_POISON calls w/o LHS.
40198 * tree-ssa.c (execute_update_addresses_taken): Create clobber
40199 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
40200 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
40201 * gimplify.c (asan_poison_variables): Add attribute
40202 use_after_scope_memory to variables that really needs to live
40204 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
40205 having the attribute.
40207 2017-01-23 Martin Liska <mliska@suse.cz>
40209 * asan.c (create_asan_shadow_var): New function.
40210 (asan_expand_poison_ifn): Likewise.
40211 * asan.h (asan_expand_poison_ifn): New declaration.
40212 * internal-fn.c (expand_ASAN_POISON): Likewise.
40213 * internal-fn.def (ASAN_POISON): New builtin.
40214 * sanopt.c (pass_sanopt::execute): Expand
40215 asan_expand_poison_ifn.
40216 * tree-inline.c (copy_decl_for_dup_finish): Make function
40218 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
40219 * tree-ssa.c (is_asan_mark_p): New function.
40220 (execute_update_addresses_taken): Rewrite local variables
40221 (identified just by use-after-scope as addressable) into SSA.
40223 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40225 * doc/install.texi (Specific): opensource.apple.com uses https
40226 now. Remove trailing slash.
40228 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40230 * README.Portability: Remove note on an Irix compatibility issue.
40232 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
40234 * gcov.c (INCLUDE_ALGORITHM): Define.
40235 (INCLUDE_VECTOR): Define.
40236 No longer include <vector> and <algorithm> directly.
40238 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
40240 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
40242 * doc/invoke.texi (Code Gen Options): Ditto.
40244 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
40247 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
40249 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
40251 rtl-optimization/79125
40252 * cprop.c (local_cprop_pass): Handle cases where we make an
40253 unconditional trap.
40255 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
40259 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
40260 read from, for big endian.
40262 2017-01-20 Jiong Wang <jiong.wang@arm.com>
40264 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
40265 register pauth builtins for LP64 only.
40267 2017-01-20 Marek Polacek <polacek@redhat.com>
40270 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
40273 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40275 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
40277 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
40278 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
40279 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
40281 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40284 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
40285 in big-endian mode when they are not a single duplicated value.
40287 2017-01-20 Richard Biener <rguenther@suse.de>
40289 * BASE-VER: Bump to 7.0.1.
40291 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40293 * omp-low.c (omplow_simd_context): New struct. Use it...
40294 (lower_rec_simd_input_clauses): ...here and...
40295 (lower_rec_input_clauses): ...here to hold common data. Adjust all
40296 references to idx, lane, max_vf, is_simt.
40298 2017-01-20 Graham Markall <graham.markall@embecosm.com>
40300 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
40303 2017-01-20 Martin Jambor <mjambor@suse.cz>
40305 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
40306 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
40308 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
40309 (GTFILES): Rename hsa.c to hsa-common.c.
40310 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
40311 * hsa-dump.c: Likewise.
40312 * hsa-gen.c: Likewise.
40313 * hsa-regalloc.c: Likewise.
40314 * ipa-hsa.c: Likewise.
40315 * omp-expand.c: Likewise.
40316 * omp-low.c: Likewise.
40317 * toplev.c: Likewise.
40319 2017-01-20 Marek Polacek <polacek@redhat.com>
40322 * doc/invoke.texi: Document -Wduplicated-branches.
40323 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
40324 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
40325 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
40326 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
40327 return 0 only when not OEP_LEXICOGRAPHIC.
40328 (fold_build_cleanup_point_expr): Use the expression
40329 location when building CLEANUP_POINT_EXPR.
40330 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
40331 * tree.c (add_expr): Handle error_mark_node.
40333 2017-01-20 Martin Liska <mliska@suse.cz>
40336 * tree-profile.c (init_ic_make_global_vars): Do not call
40338 (gimple_init_gcov_profiler): Likewise.
40340 2017-01-20 Martin Liska <mliska@suse.cz>
40343 * cgraph.h (maybe_create_reference): Remove argument and
40345 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
40347 * ipa-cp.c (create_specialized_node): Likewise.
40348 * symtab.c (symtab_node::maybe_create_reference): Handle
40349 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
40351 2017-01-20 Martin Liska <mliska@suse.cz>
40353 * read-rtl-function.c (function_reader::create_function): Use
40354 build_decl instread of build_decl_stat.
40356 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
40358 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
40359 * config/i386/avx512dqintrin.h: Ditto.
40360 * config/i386/avx512fintrin.h: Ditto.
40361 * config/i386/i386-builtin-types.def: Add new types.
40362 * config/i386/i386.c: Handle new types.
40363 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
40364 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
40365 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
40366 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
40367 (__builtin_ia32_kshiftridi): New.
40368 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
40370 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
40374 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
40375 define to rs6000_init_stack_protect_guard.
40376 (rs6000_init_stack_protect_guard): New function.
40378 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40379 Yunqiang Su <yunqiang.su@imgtec.com>
40381 * config.gcc (supported_defaults): Add madd4.
40382 (with_madd4): Add validation.
40383 (all_defaults): Add madd4.
40384 * config/mips/mips.opt (mmadd4): New option.
40385 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40387 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40389 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
40390 (ISA_HAS_FUSED_MADD4): Likewise.
40391 * doc/invoke.texi (-mmadd4): Document the new option.
40392 * doc/install.texi (--with-madd4): Document the new option.
40394 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40396 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
40397 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
40398 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
40399 (aarch64_init_pauth_hint_builtins): New.
40400 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
40401 (aarch64_expand_builtin): Expand new builtins.
40403 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40405 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
40406 * combine-stack-adj.c (no_unhandled_cfa): Handle
40407 REG_CFA_TOGGLE_RA_MANGLE.
40408 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
40409 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
40410 info for return address signing.
40411 (aarch64_expand_epilogue): Likewise.
40413 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40415 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
40416 * config/aarch64/aarch64-protos.h
40417 (aarch64_return_address_signing_enabled): New declaration.
40418 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
40420 (aarch64_expand_prologue): Sign return address before it's pushed onto
40422 (aarch64_expand_epilogue): Authenticate return address fetched from
40424 (aarch64_override_options): Sanity check for ILP32 and ISA level.
40425 (aarch64_attributes): New function attributes for "sign-return-address".
40426 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
40427 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
40428 ("*do_return"): Generate combined instructions according to key index.
40429 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
40430 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
40432 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
40433 * config/aarch64/aarch64.opt (msign-return-address=): New.
40434 * doc/extend.texi (AArch64 Function Attributes): Documents
40435 "sign-return-address=".
40436 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
40438 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40440 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
40441 overall option summary.
40443 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40445 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
40446 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
40447 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
40448 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
40450 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
40452 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
40453 -mpower9-minmax by default for -mcpu=power9.
40454 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
40455 128-bit floating point.
40457 2017-01-20 Alan Modra <amodra@gmail.com>
40459 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
40460 optimizing for size.
40462 2017-01-20 Alan Modra <amodra@gmail.com>
40465 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
40466 for strcmp and strncmp from corresponding builtin decl.
40468 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40470 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
40471 instead of i386/rtems-64.h.
40472 * config/i386/rtems-64.h: Remove.
40474 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40478 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
40480 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
40482 2017-01-19 Tamar Christina <tamar.christina@arm.com>
40484 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
40485 Change int to HOST_WIDE_INT.
40486 * config/aarch64/aarch64-protos.h
40487 (aarch64_simd_gen_const_vector_dup): Likewise.
40488 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
40490 2017-01-19 David Malcolm <dmalcolm@redhat.com>
40492 * langhooks-def.h (lhd_type_for_size): New decl.
40493 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
40494 * langhooks.c (lhd_type_for_size): New function, taken from
40497 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
40499 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
40500 define_bypass for CR latency.
40501 (power9-cracked-alu): Update bypass latency and remove power9-branch.
40502 (power9-alu2): Add define_bypass for CR latency.
40504 (power9-mul): Update insn latency.
40505 (power9-mul-compare): Update insn latency, bypass latency and remove
40508 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40510 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
40512 * config/aarch64/aarch64.md
40513 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
40514 aarch64_nopcrelative_literal_loads.
40515 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
40517 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
40519 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
40520 TARGET_LOONGSON_3A.
40521 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
40523 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
40526 * config.gcc (supported_defaults): Add lxc1-sxc1.
40527 (with_lxc1_sxc1): Add validation.
40528 (all_defaults): Add lxc1-sxc1.
40529 * config/mips/mips.opt (mlxc1-sxc1): New option.
40530 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40532 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40533 __mips_no_lxc1_sxc1.
40534 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
40535 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
40536 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
40538 2017-01-19 Richard Biener <rguenther@suse.de>
40540 PR tree-optimization/72488
40541 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
40542 sure to restore SSA info.
40543 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
40545 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
40547 PR rtl-optimization/79121
40548 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
40549 of the inner type when shifting an extended value.
40551 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40554 * symtab.c (symtab_node::equal_address_to): Fix comparing of
40555 interposable aliases.
40557 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
40560 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
40561 Use the evmergelohi instruction.
40562 (mov_si<mode>_e500_subreg4_2_le): Likewise.
40563 (mov_sitf_e500_subreg8_2_be): Likewise.
40564 (mov_sitf_e500_subreg12_2_le): Likewise.
40565 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
40566 (mov_si<mode>_e500_subreg4_2_be): Likewise.
40567 (mov_sitf_e500_subreg8_2_le): Likewise.
40568 (mov_sitf_e500_subreg12_2_be): Likewise.
40570 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40572 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
40573 attribute from vecsimple to vecperm.
40574 (altivec_vbpermq2): Likewise.
40576 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40579 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
40581 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
40582 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
40583 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
40584 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
40585 case where N arg is SIZE_MAX.
40586 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
40587 (cmpstrsi): Add pattern.
40589 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
40591 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
40592 __builtin_vec_revb builtins.
40593 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
40594 built-in functions to support generation of the ISA 3.0 XXBR<x>
40595 vector byte reverse instructions.
40596 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
40597 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
40598 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
40599 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
40600 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
40601 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
40602 (P9V_BUILTIN_VEC_REVB): Likewise.
40603 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
40604 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
40605 (p9_xxbrq_v16qi): Likewise.
40606 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
40607 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
40608 (p9_xxbrh_v8hi): Likewise.
40609 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
40610 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
40611 vec_revb built-in functions.
40613 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
40615 PR rtl-optimization/78952
40616 * config/i386/i386.md (any_extract): New code iterator.
40617 (*insvqi_2): Use any_extract for source operand.
40618 (*insvqi_3): Use any_shiftrt for source operand.
40620 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
40622 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
40624 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
40626 2017-01-18 Matthias Klose <doko@ubuntu.com>
40628 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
40630 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40632 * config/rs6000/altivec.h (vec_bperm): Change #define.
40633 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
40634 (altivec_vbpermq2): New define_insn.
40635 (altivec_vbpermd): Likewise.
40636 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
40637 function interface.
40638 (VBPERMD): Likewise.
40639 (VBPERM): New polymorphic function interface.
40640 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
40641 Add entries for P9V_BUILTIN_VEC_VBPERM.
40642 * doc/extend.texi: Add interfaces for vec_bperm.
40644 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40646 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
40647 first letter of error messages.
40648 (s390_resolve_overloaded_builtin): Likewise.
40649 * config/s390/s390.c (s390_expand_builtin): Likewise.
40650 (s390_invalid_arg_for_unprototyped_fn): Likewise.
40651 (s390_valid_target_attribute_inner_p): Likewise.
40652 * config/s390/s390.md ("tabort"): Likewise.
40654 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
40656 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
40657 (ISA_AVOID_DIV_HILO): New macro.
40658 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
40659 (ISA_HAS_DDIV): Likewise.
40661 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40663 * doc/invoke.texi (fabi-version): Correct number of occurrences.
40665 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40667 * doc/invoke.texi (fabi-version): Spelling fix.
40669 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40672 * doc/invoke.texi (fabi-version): Mention mangling fix for
40675 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40678 * doc/invoke.texi (fabi-version): Document discriminator mangling.
40680 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
40683 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
40684 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
40686 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
40688 (stack_protect_test): Ditto.
40689 * config/rs6000/rs6000.opt (mstack-protector-guard=,
40690 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
40692 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
40693 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
40694 -mstack-protector-guard-offset=.
40695 (RS/6000 and PowerPC Options): Ditto.
40697 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40699 * config/i386/i386.h (MASK_CLASS_P): New define.
40700 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
40701 there are no registers from different register sets also when
40702 mask registers are used. Update function comment.
40703 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
40704 to (*k/*r) and (*k/*km) alternatives.
40706 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
40708 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
40709 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
40710 (EH_RETURN_HANDLER_RTX): New define.
40711 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
40712 Force frame pointer in EH return functions.
40713 (aarch64_expand_epilogue): Add barrier for eh_return.
40714 (aarch64_final_eh_return_addr): Remove.
40715 (aarch64_eh_return_handler_rtx): New function.
40716 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
40718 (aarch64_eh_return_handler_rtx): New prototype.
40720 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40722 * config/rs6000/altivec.h (vec_rlmi): New #define.
40723 (vec_vrlnm): Likewise.
40724 (vec_rlnm): Likewise.
40725 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
40726 (UNSPEC_VRLNM): Likewise.
40727 (VIlong): New mode iterator.
40728 (altivec_vrl<VI_char>mi): New define_insn.
40729 (altivec_vrl<VI_char>nm): Likewise.
40730 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
40732 (VRLDNM): Likewise.
40733 (RLNM): New polymorphic function entry.
40734 (VRLWMI): New monomorphic function entry.
40735 (VRLDMI): Likewise.
40736 (RLMI): New polymorphic function entry.
40737 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
40738 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
40739 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
40742 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40745 * dwarf2out.c (field_byte_offset): Restore the
40746 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
40747 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
40748 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
40751 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40754 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
40756 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40759 * dwarf2out.c (add_data_member_location_attribute): For constant
40760 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
40761 instead of DW_AT_data_member_location, DW_AT_bit_offset and
40762 DW_AT_byte_size attributes.
40764 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40766 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
40767 after forcing to constant memory when the code model is medium.
40769 2017-01-17 Julia Koval <julia.koval@intel.com>
40772 * config/i386/avx512fintrin.h
40773 (_mm512_i32gather_ps): Change __addr type to void const*.
40774 (_mm512_mask_i32gather_ps): Ditto.
40775 (_mm512_i32gather_pd): Ditto.
40776 (_mm512_mask_i32gather_pd): Ditto.
40777 (_mm512_i64gather_ps): Ditto.
40778 (_mm512_mask_i64gather_ps): Ditto.
40779 (_mm512_i64gather_pd): Ditto.
40780 (_mm512_mask_i64gather_pd): Ditto.
40781 (_mm512_i32gather_epi32): Ditto.
40782 (_mm512_mask_i32gather_epi32): Ditto.
40783 (_mm512_i32gather_epi64): Ditto.
40784 (_mm512_mask_i32gather_epi64): Ditto.
40785 (_mm512_i64gather_epi32): Ditto.
40786 (_mm512_mask_i64gather_epi32): Ditto.
40787 (_mm512_i64gather_epi64): Ditto.
40788 (_mm512_mask_i64gather_epi64): Ditto.
40789 (_mm512_i32scatter_ps): Change __addr type to void*.
40790 (_mm512_mask_i32scatter_ps): Ditto.
40791 (_mm512_i32scatter_pd): Ditto.
40792 (_mm512_mask_i32scatter_pd): Ditto.
40793 (_mm512_i64scatter_ps): Ditto.
40794 (_mm512_mask_i64scatter_ps): Ditto.
40795 (_mm512_i64scatter_pd): Ditto.
40796 (_mm512_mask_i64scatter_pd): Ditto.
40797 (_mm512_i32scatter_epi32): Ditto.
40798 (_mm512_mask_i32scatter_epi32): Ditto.
40799 (_mm512_i32scatter_epi64): Ditto.
40800 (_mm512_mask_i32scatter_epi64): Ditto.
40801 (_mm512_i64scatter_epi32): Ditto.
40802 (_mm512_mask_i64scatter_epi32): Ditto.
40803 (_mm512_i64scatter_epi64): Ditto.
40804 (_mm512_mask_i64scatter_epi64): Ditto.
40805 * config/i386/avx512pfintrin.h
40806 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40807 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40808 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40809 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40810 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40811 (_mm512_prefetch_i32scatter_ps): Ditto.
40812 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40813 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40814 (_mm512_prefetch_i64scatter_pd): Ditto.
40815 (_mm512_prefetch_i64scatter_ps): Ditto.
40816 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40817 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40818 * config/i386/avx512vlintrin.h
40819 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40820 (_mm_mmask_i32gather_ps): Ditto.
40821 (_mm256_mmask_i32gather_pd): Ditto.
40822 (_mm_mmask_i32gather_pd): Ditto.
40823 (_mm256_mmask_i64gather_ps): Ditto.
40824 (_mm_mmask_i64gather_ps): Ditto.
40825 (_mm256_mmask_i64gather_pd): Ditto.
40826 (_mm_mmask_i64gather_pd): Ditto.
40827 (_mm256_mmask_i32gather_epi32): Ditto.
40828 (_mm_mmask_i32gather_epi32): Ditto.
40829 (_mm256_mmask_i32gather_epi64): Ditto.
40830 (_mm_mmask_i32gather_epi64): Ditto.
40831 (_mm256_mmask_i64gather_epi32): Ditto.
40832 (_mm_mmask_i64gather_epi32): Ditto.
40833 (_mm256_mmask_i64gather_epi64): Ditto.
40834 (_mm_mmask_i64gather_epi64): Ditto.
40835 (_mm256_i32scatter_ps): Change __addr type to void*.
40836 (_mm256_mask_i32scatter_ps): Ditto.
40837 (_mm_i32scatter_ps): Ditto.
40838 (_mm_mask_i32scatter_ps): Ditto.
40839 (_mm256_i32scatter_pd): Ditto.
40840 (_mm256_mask_i32scatter_pd): Ditto.
40841 (_mm_i32scatter_pd): Ditto.
40842 (_mm_mask_i32scatter_pd): Ditto.
40843 (_mm256_i64scatter_ps): Ditto.
40844 (_mm256_mask_i64scatter_ps): Ditto.
40845 (_mm_i64scatter_ps): Ditto.
40846 (_mm_mask_i64scatter_ps): Ditto.
40847 (_mm256_i64scatter_pd): Ditto.
40848 (_mm256_mask_i64scatter_pd): Ditto.
40849 (_mm_i64scatter_pd): Ditto.
40850 (_mm_mask_i64scatter_pd): Ditto.
40851 (_mm256_i32scatter_epi32): Ditto.
40852 (_mm256_mask_i32scatter_epi32): Ditto.
40853 (_mm_i32scatter_epi32): Ditto.
40854 (_mm_mask_i32scatter_epi32): Ditto.
40855 (_mm256_i32scatter_epi64): Ditto.
40856 (_mm256_mask_i32scatter_epi64): Ditto.
40857 (_mm_i32scatter_epi64): Ditto.
40858 (_mm_mask_i32scatter_epi64): Ditto.
40859 (_mm256_i64scatter_epi32): Ditto.
40860 (_mm256_mask_i64scatter_epi32): Ditto.
40861 (_mm_i64scatter_epi32): Ditto.
40862 (_mm_mask_i64scatter_epi32): Ditto.
40863 (_mm256_i64scatter_epi64): Ditto.
40864 (_mm256_mask_i64scatter_epi64): Ditto.
40865 (_mm_i64scatter_epi64): Ditto.
40866 (_mm_mask_i64scatter_epi64): Ditto.
40867 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40868 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40869 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40870 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40871 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40872 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40873 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40874 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40875 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40876 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40877 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40878 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40879 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40880 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40881 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40882 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40883 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40884 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40885 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40886 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40887 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40888 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40889 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40890 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40891 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40892 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40893 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40894 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40895 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40896 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40897 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40898 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40899 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40900 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40901 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40902 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40903 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40904 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40905 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40906 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40907 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40908 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40909 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40910 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40911 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
40912 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
40913 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
40914 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
40915 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
40916 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
40917 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
40918 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
40919 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
40920 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
40921 definitions accordingly.
40923 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
40924 Kuan-Lin Chen <kuanlinchentw@gmail.com>
40927 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
40930 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
40933 * ira-conflicts.c (ira_build_conflicts): Update total conflict
40934 hard regs for inner regno.
40936 2017-01-17 Martin Liska <mliska@suse.cz>
40939 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
40940 assumption and add comment.
40942 2017-01-17 Nathan Sidwell <nathan@acm.org>
40944 * ipa-visibility.c (localize_node): New function, broken out of ...
40945 (function_and_variable_visibility): ... here. Call it.
40947 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40949 PR middle-end/77445
40950 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
40951 correctly set frequency of oudgoing edge.
40952 (duplicate_thread_path): Fix profile updating.
40954 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40957 * configure.ac: Add GCC_BASE_VER.
40958 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
40959 version from BASE-VER file.
40960 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
40961 (gcc.o): Depend on $(BASEVER).
40962 * common.opt (dumpfullversion): New option.
40963 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
40964 * doc/invoke.texi: Document -dumpfullversion.
40965 * doc/install.texi: Document --with-gcc-major-version-only.
40966 * configure: Regenerated.
40968 2017-01-17 Richard Biener <rguenther@suse.de>
40970 PR tree-optimization/71433
40971 * tree-vrp.c (register_new_assert_for): Merge same asserts
40972 on all incoming edges.
40973 (process_assert_insertions_for): Handle insertions at the
40976 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
40978 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
40979 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
40981 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
40984 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
40987 2017-01-17 Alan Modra <amodra@gmail.com>
40990 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
40991 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
40992 symbolic stack limit when pic.
40994 2017-01-16 Martin Sebor <msebor@redhat.com>
40996 PR tree-optimization/78608
40997 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
40999 2017-01-16 Jeff Law <law@redhat.com>
41002 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
41003 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
41004 for several include directories that may be relative to sysroot.
41005 * config/i386/x-mingw32 (gplus_includedir): Define.
41006 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
41007 (native_system_includedir): Likewise.
41008 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
41009 override if TARGET_SYSTEM_ROOT is defined.
41010 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
41012 PR tree-optimization/79090
41013 PR tree-optimization/33562
41014 PR tree-optimization/61912
41015 PR tree-optimization/77485
41016 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
41017 and computed trims into the dump file.
41019 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
41021 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
41023 2017-01-16 Jakub Jelinek <jakub@redhat.com>
41026 * gimplify.c (gimplify_init_constructor): If want_value and
41027 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
41031 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
41032 sequence. Formatting fixes.
41033 (doloop_optimize): Formatting fixes.
41036 * gcc.c (debug_level_greater_than_spec_func): New function.
41037 (static_spec_functions): Add debug-level-gt spec function.
41038 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
41040 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
41041 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
41042 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
41043 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
41044 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
41045 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
41047 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
41049 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
41050 QImode fixups to general and mask registers only.
41052 2017-01-16 Carl Love <cel@us.ibm.com>
41054 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
41055 for built-in functions
41056 vector signed char vec_nabs (vector signed char)
41057 vector signed short vec_nabs (vector signed short)
41058 vector signed int vec_nabs (vector signed int)
41059 vector signed long long vec_nabs (vector signed long long)
41060 vector float vec_nabs (vector float)
41061 vector double vec_nabs (vector double)
41062 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
41064 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
41065 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
41066 * doc/extend.texi: Update the documentation file for the new built-in
41069 2017-01-16 Martin Sebor <msebor@redhat.com>
41071 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
41074 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41076 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
41077 UNSPEC_VSX__XXSPLTD to require special splat handling.
41079 2017-01-16 David Malcolm <dmalcolm@redhat.com>
41082 * system.h: Poison strndup.
41084 2017-01-16 Alan Modra <amodra@gmail.com>
41087 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
41090 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
41092 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
41094 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
41096 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
41097 call recog here. Assert that INSN_CODE (insn) is non-negative.
41099 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41102 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
41104 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
41105 in the currently scheduled RTL fragment.
41107 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41109 PR rtl-optimization/78751
41110 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
41113 2017-01-14 Jeff Law <law@redhat.com>
41115 PR tree-optimization/79090
41116 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
41117 variable length stores.
41118 (compute_trims): Delete dead assignment to *trim_tail.
41119 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
41122 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
41124 PR rtl-optimization/78626
41125 PR rtl-optimization/78727
41126 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
41127 of a block, and split such blocks after everything else is finished.
41129 2017-01-14 Alan Modra <amodra@gmail.com>
41132 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
41133 target legitimate_combined_insn.
41134 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
41135 (rs6000_legitimate_combined_insn): New function.
41136 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
41138 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
41139 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
41140 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
41142 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
41144 * doc/frontends.texi (G++ and GCC): Remove references to Java.
41146 2017-01-13 Jeff Law <law@redhat.com>
41148 PR tree-optimization/33562
41149 PR tree-optimization/61912
41150 PR tree-optimization/77485
41151 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
41153 (delete_dead_assignment): Likewise.
41154 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
41155 statement to delete_dead_call and delete_dead_assignment.
41157 2017-01-13 David Malcolm <dmalcolm@redhat.com>
41160 * substring-locations.c (format_warning_va): Strengthen case 1 so
41161 that both endpoints of the substring must be within the format
41162 range for just the substring to be printed.
41164 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
41166 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
41167 * config/i386/i386.c (ix86_target_string): Add missing options
41168 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
41169 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
41170 flags_other and ix86_target_other to flags2_other. Display unknown
41172 (ix86_valid_target_attribute_inner_p): Add missing options and
41173 reorder options by implied ISAs, as in ix86_target_string.
41175 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41177 * hash-table.h (hash_table::too_empty_p): New function.
41178 (hash_table::expand): Use it.
41179 (hash_table::traverse): Likewise.
41180 (hash_table::empty_slot): Use sizeof (value_type) instead of
41181 sizeof (PTR) to convert bytes to elements. Shrink the table
41182 if the current size is excessive for the current number of
41185 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41187 * ira-costs.c (record_reg_classes): Break from the inner loop
41188 early once alt_fail is known to be true. Update outer loop
41189 handling accordingly.
41191 2017-01-13 Jeff Law <law@redhat.com>
41193 * tree-ssa-dse.c (decrement_count): New function.
41194 (increment_start_addr, maybe_trim_memstar_call): Likewise.
41195 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
41196 when we know the partially dead statement is a mem* function.
41198 PR tree-optimization/61912
41199 PR tree-optimization/77485
41200 * tree-ssa-dse.c: Include expr.h.
41201 (maybe_trim_constructor_store): New function.
41202 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
41204 PR tree-optimization/33562
41205 PR tree-optimization/61912
41206 PR tree-optimization/77485
41207 * doc/invoke.texi: Document new dse-max-object-size param.
41208 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
41209 * tree-ssa-dse.c: Include params.h.
41210 (dse_store_status): New enum.
41211 (initialize_ao_ref_for_dse): New, partially extracted from
41213 (valid_ao_ref_for_dse, normalize_ref): New.
41214 (setup_live_bytes_from_ref, compute_trims): Likewise.
41215 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
41216 (maybe_trim_partially_dead_store): Likewise.
41217 (maybe_trim_complex_store): Likewise.
41218 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
41219 Track what bytes live from the original store. Return tri-state
41220 for dead, partially dead or live.
41221 (dse_dom_walker): Add constructor, destructor and new private members.
41222 (delete_dead_call, delete_dead_assignment): New extracted from
41224 (dse_optimize_stmt): Make a member of dse_dom_walker.
41225 Use initialize_ao_ref_for_dse.
41227 PR tree-optimization/33562
41228 PR tree-optimization/61912
41229 PR tree-optimization/77485
41230 * sbitmap.h (bitmap_count_bits): Prototype.
41231 (bitmap_clear_range, bitmap_set_range): Likewise.
41232 * sbitmap.c (bitmap_clear_range): New function.
41233 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
41235 2017-01-13 Martin Liska <mliska@suse.cz>
41238 * function.c (set_cfun): Add new argument force.
41239 * function.h (set_cfun): Likewise.
41240 * ipa-inline-transform.c (inline_call): Use the function when
41241 strict alising from is dropped for function we inline to.
41243 2017-01-13 Richard Biener <rguenther@suse.de>
41245 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
41246 for dumping GIMPLE INTEGER_CSTs.
41248 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
41250 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
41251 to 201112L since C++17.
41253 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
41256 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
41257 if -fsanitize=kernel-address is present.
41259 2017-01-13 Richard Biener <rguenther@suse.de>
41261 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
41262 as _Literal ( type ) number in case usual suffixes do not
41263 preserve all information.
41265 2017-01-13 Richard Biener <rguenther@suse.de>
41267 PR tree-optimization/77283
41268 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
41269 and ssa-iterators.h.
41270 (is_feasible_trace): Implement a cost model based on joiner
41273 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
41276 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
41277 char or short to __float128/_Float128 directly.
41279 2017-01-12 Martin Sebor <msebor@redhat.com>
41281 to -Wformat-overflow.
41282 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
41283 (min_bytes_remaining): Same.
41284 (get_string_length): Same.
41285 (format_string): Same.
41286 (format_directive): Same.
41288 (pass_sprintf_length::handle_gimple_call): Same.
41290 2017-01-12 Jakub Jelinek <jakub@redhat.com>
41292 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
41293 info.nowrite calls with no lhs that can't throw. Return bool
41294 whether gsi_remove has been called or not.
41295 (pass_sprintf_length::handle_gimple_call): Return bool whether
41296 try_substitute_return_value called gsi_remove. Formatting fix.
41297 (pass_sprintf_length::execute): Don't use gsi_remove if
41298 handle_gimple_call returned true.
41301 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
41302 be removed due to side-effects, don't remove following barrier nor
41303 turn the successor edge into fallthru edge.
41305 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41308 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
41309 element-reversing loads and stores as not swappable.
41311 2017-01-12 Nathan Sidwell <nathan@acm.org>
41312 Nicolai Stange <nicstange@gmail.com>
41314 * combine.c (try_combine): Don't ignore result of overlap checking
41315 loop. Combine overlap & asm check into single loop.
41317 2017-01-12 Richard Biener <rguenther@suse.de>
41319 * tree-pretty-print.c (dump_generic_node): Provide -gimple
41320 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
41322 2017-01-12 Richard Biener <rguenther@suse.de>
41324 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
41325 and TS_TARGET_OPTION directly derive from TS_BASE.
41326 * tree-core.h (tree_optimization_option): Derive from tree_base.
41327 (tree_target_option): Likewise.
41329 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
41331 * config/i386/i386.c (memory_address_length): Increase len
41332 only when rip_relative_addr_p returns false.
41334 2017-01-11 Julia Koval <julia.koval@intel.com>
41336 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
41337 (OPTION_MASK_ISA_SGX_SET): New.
41338 (ix86_handle_option): Handle OPT_msgx.
41339 * config.gcc: Added sgxintrin.h.
41340 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
41341 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
41342 * config/i386/i386.c (ix86_target_string): Add -msgx.
41344 (ix86_option_override_internal): Handle new options.
41345 (ix86_valid_target_attribute_inner_p): Add sgx.
41346 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
41347 * config/i386/i386.opt: Add msgx.
41348 * config/i386/sgxintrin.h: New file.
41349 * config/i386/x86intrin.h: Add sgxintrin.h.
41351 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41354 * fold-const.c (maybe_nonzero_address): Return 1 for function
41356 (tree_single_nonzero_warnv_p): Don't handle function local objects
41360 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
41363 2017-01-11 David Malcolm <dmalcolm@redhat.com>
41366 * opts.c: Include "spellcheck.h"
41367 (struct string_fragment): New struct.
41368 (struct edit_distance_traits<const string_fragment &>): New
41370 (get_closest_sanitizer_option): New function.
41371 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
41373 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41375 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
41377 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
41378 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
41379 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
41380 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
41381 for initial die_offset if dwarf_split_debug_info.
41382 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
41383 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
41385 (output_skeleton_debug_sections): Formatting fix. Use
41386 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
41387 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
41389 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
41391 * config/arm/cortex-a53.md: Add bypasses for
41392 cortex_a53_r2f_cvt.
41393 (cortex_a53_r2f): Only use for transfers.
41394 (cortex_a53_f2r): Likewise.
41395 (cortex_a53_r2f_cvt): Add reservation for conversions.
41396 (cortex_a53_f2r_cvt): Likewise.
41398 2017-01-11 Tamar Christina <tamar.christina@arm.com>
41400 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
41401 to all inlined functions, change static to extern.
41403 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
41406 * config/arm/arm.c (legitimize_pic_address): Handle reference to
41408 (arm_assemble_integer): Likewise.
41410 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
41412 * config.gcc: Use new awk script to check CPU, FPU and architecture
41413 parameters for --with-... options.
41414 * config/arm/parsecpu.awk: New file
41415 * config/arm/arm-cpus.in: New file.
41416 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
41418 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
41420 * config/arm/t-arm: Update dependency rules.
41421 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
41422 of processing .def files.
41423 * config/arm/genopt.sh: Deleted.
41424 * config/arm/gentune.sh: Deleted.
41425 * config/arm/arm-cores.def: Deleted.
41426 * config/arm/arm-arches.def: Deleted.
41427 * config/arm/arm-fpus.def: Deleted.
41428 * config/arm/arm-tune.md: Regenerated.
41429 * config/arm/arm-tables.opt: Regenerated.
41430 * config/arm/arm-cpu.h: New generated file.
41431 * config/arm/arm-cpu-data.h: New generated file.
41432 * config/arm/arm-cpu-cdata.h: New generated file.
41434 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
41437 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
41439 (input_varpool_node): Unpack dynamically_initialized bit.
41441 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
41443 PR rtl-optimization/79032
41444 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
41445 the alignment of the adjusted memory reference against that of MODE,
41446 instead of the alignment of the original memory reference.
41448 2017-01-11 Martin Jambor <mjambor@suse.cz>
41450 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
41452 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
41453 decorated functions.
41455 2017-01-11 Richard Biener <rguenther@suse.de>
41457 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
41458 set range/nonnull info for PHI results. Do not set it on
41459 stmts marked for removal.
41461 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
41463 * expr.c (store_field): In the bitfield case, fetch the return value
41464 from the registers before applying a single big-endian adjustment.
41465 Always do a final load for a BLKmode value not larger than a word.
41467 2017-01-10 David Malcolm <dmalcolm@redhat.com>
41470 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41471 that we correctly handle column numbers greater than
41472 LINE_MAP_MAX_COLUMN_NUMBER.
41474 2017-01-10 Martin Sebor <msebor@redhat.com>
41476 PR middle-end/78245
41477 * gimple-ssa-sprintf.c (get_destination_size): Call
41478 {init,fini}object_sizes.
41479 * tree-object-size.c (addr_object_size): Adjust.
41480 (pass_through_call): Adjust.
41481 (pass_object_sizes::execute): Adjust.
41482 * tree-object-size.h (fini_object_sizes): Declare.
41484 2017-01-10 Martin Sebor <msebor@redhat.com>
41486 PR tree-optimization/78775
41487 * builtins.c (get_size_range): Move...
41488 * calls.c: ...to here.
41489 (alloc_max_size): Accept zero argument.
41490 (operand_signed_p): Remove.
41491 (maybe_warn_alloc_args_overflow): Call get_size_range.
41492 * calls.h (get_size_range): Declare.
41494 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
41496 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
41497 from TI's devices.csv file as of September 2016.
41498 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
41500 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
41502 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
41503 * doc/invoke.texi: Likewise.
41504 * doc/md.texi: Likewise.
41505 * doc/objc.texi: Likewise.
41507 2017-01-10 Joshua Conner <joshconner@google.com>
41509 * config/arm/fuchsia-elf.h: New file.
41510 * config/fuchsia.h: New file.
41511 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
41512 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
41514 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
41516 2016-01-10 Richard Biener <rguenther@suse.de>
41518 PR tree-optimization/79034
41519 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
41520 Propagate out degenerate PHIs in the joiner.
41522 2017-01-10 Martin Liska <mliska@suse.cz>
41524 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
41525 (sort_congruence_classes_by_decl_uid): Likewise.
41526 (sort_congruence_class_groups_by_decl_uid): Likewise.
41527 (sem_item_optimizer::merge_classes): Sort class, groups in these
41528 classes and members in the groups by DECL_UID of declarations.
41529 This would make merge operations stable.
41531 2017-01-10 Martin Liska <mliska@suse.cz>
41533 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
41534 usage of m_classes_vec.
41535 (sem_item_optimizer::~sem_item_optimizer): Likewise.
41536 (sem_item_optimizer::get_group_by_hash): Likewise.
41537 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
41538 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
41539 (sem_item_optimizer::verify_classes): Likewise.
41540 (sem_item_optimizer::process_cong_reduction): Likewise.
41541 (sem_item_optimizer::dump_cong_classes): Likewise.
41542 (sem_item_optimizer::merge_classes): Likewise.
41543 * ipa-icf.h (congruence_class_hash): Rename from
41544 congruence_class_group_hash. Remove declaration of m_classes_vec.
41546 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
41548 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
41549 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
41550 * config.gcc: Add avx512vpopcntdqintrin.h.
41551 * config/i386/avx512vpopcntdqintrin.h: New.
41552 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
41553 * config/i386/i386-builtin-types.def: Add new types.
41554 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
41555 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
41556 __builtin_ia32_vpopcountq_v8di_mask): New.
41557 * config/i386/i386-c.c (ix86_target_macros_internal): Define
41558 __AVX512VPOPCNTDQ__.
41559 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
41560 (PTA_AVX512VPOPCNTDQ): Define.
41561 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
41562 TARGET_AVX512VPOPCNTDQ_P): Define.
41563 * config/i386/i386.opt: Add mavx512vpopcntdq.
41564 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
41565 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
41567 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41569 PR middle-end/77484
41570 * predict.def (PRED_CALL): Set to 67.
41572 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
41574 * expr.c (store_field): In the bitfield case, if the value comes from
41575 a function call and is of an aggregate type returned in registers, do
41576 not modify the field mode; extract the value in all cases if the mode
41577 is BLKmode and the size is not larger than a word.
41579 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
41582 * config/i386/cpuid.h: Fix undefined behavior.
41584 2017-01-04 Jeff Law <law@redhat.com>
41586 PR tree-optimization/79007
41587 PR tree-optimization/67955
41588 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
41589 conservative for pt.null when flag_non_call_exceptions is on.
41591 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41593 PR translation/79019
41594 PR translation/79020
41595 * params.def (PARAM_INLINE_MIN_SPEEDUP,
41596 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
41597 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
41599 * config/avr/avr.opt (maccumulate-args): Likewise.
41600 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
41601 * common.opt (freport-bug): Likewise.
41602 * cif-code.def (CIF_FINAL_ERROR): Likewise.
41603 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
41604 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
41605 translatable string.
41606 * config/i386/i386.c (function_value_32): Likewise.
41607 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
41608 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
41610 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
41611 * common/config/msp430/msp430-common.c (msp430_handle_option):
41613 * symtab.c (symtab_node::verify_base): Likewise.
41614 * opts.c (set_debug_level): Likewise.
41615 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
41616 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
41617 missing whitespace to translatable strings.
41618 * config/avr/avr.md (bswapsi2): Fix typo in comment.
41619 * config/sh/superh.h: Likewise.
41620 * config/i386/xopintrin.h: Likewise.
41621 * config/i386/znver1.md: Likewise.
41622 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
41623 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
41624 * double-int.h (struct double_int): Likewise.
41625 * double-int.c (div_and_round_double): Likewise.
41626 * wide-int.cc: Likewise.
41627 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
41628 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
41629 * cfgcleanup.c (crossjumps_occured): Renamed to ...
41630 (crossjumps_occurred): ... this.
41631 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
41634 PR tree-optimization/78899
41635 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
41636 returning bool return struct loop *, NULL for failure and the new
41638 (versionable_outer_loop_p): Don't version outer loop if it has
41639 dont_vectorized bit set.
41640 (tree_if_conversion): When versioning outer loop, ensure
41641 tree_if_conversion is performed also on the inner loop of the
41642 non-vectorizable outer loop copy.
41643 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
41644 LOOP_VECTORIZED in inner loop of the scalar outer loop and
41645 prevent vectorization of it.
41646 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
41647 the outer loop vectorization of the non-scalar version is attempted
41648 before vectorization of the inner loop in scalar version. If
41649 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
41650 vectorization of its inner loop.
41651 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
41652 has 2 inner loops, rename also on edges from bb whose single pred
41653 is outer_loop->header. Fix typo in function comment.
41655 2017-01-09 Martin Sebor <msebor@redhat.com>
41658 * asan.c (asan_emit_stack_protection): Increase local buffer size
41659 to avoid snprintf truncation warning.
41661 2017-01-09 Andrew Pinski <apinski@cavium.com>
41663 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
41664 to reference thunderx2t99 for the tuning structure
41665 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
41667 (thunderx2t99_extra_costs): This.
41668 * config/aarch64/aarch64-tune.md: Regenerate.
41669 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
41670 (vulcan_addrcost_table): This.
41671 (vulcan_regmove_cost): Rename to ...
41672 (thunderx2t99_regmove_cost): This.
41673 (vulcan_vector_cost): Rename to ...
41674 (thunderx2t99_vector_cost): this.
41675 (vulcan_branch_cost): Rename to ...
41676 (thunderx2t99_branch_cost): This.
41677 (vulcan_tunings): Rename to ...
41678 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
41679 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
41681 2017-01-09 Martin Jambor <mjambor@suse.cz>
41685 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
41686 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
41687 (propagate_vr_accross_jump_function): Use the above function for all
41688 value range computations for pass-through jump functions and type
41689 converasion from explicit value range values.
41690 (ipcp_propagate_stage): Do not attempt to deduce types of formal
41691 parameters from TYPE_ARG_TYPES.
41692 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
41693 (ipa_write_node_info): Stream type of the actual argument.
41694 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
41696 2017-01-09 Martin Liska <mliska@suse.cz>
41699 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
41700 (lookup_compiler): Do not show error message with have_E.
41702 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41704 PR tree-optimization/78938
41705 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
41706 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
41707 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
41708 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
41711 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
41713 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
41716 2017-01-09 Richard Biener <rguenther@suse.de>
41718 PR tree-optimization/78997
41719 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
41720 name condition properly.
41722 2017-01-09 Richard Biener <rguenther@suse.de>
41725 * dwarf2out.c (is_cxx): New overload with context.
41726 (is_naming_typedef_decl): Use it.
41728 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41730 * invoke.texi (Option Summary): Correct spacing in option lists
41731 and add line breaks to fix over-long lines.
41733 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41735 PR middle-end/17660
41737 * extend.texi (Common Variable Attributes): Add xref to GCC
41738 Internals manual to explain mode attribute keywords.
41740 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41743 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
41744 and Preprocessor Options.
41745 (Options for Linking): Document -pthread here....
41746 (RS/6000 and PowerPC Options): ...not here.
41747 (Solaris 2 Options): ...or here.
41748 * doc/cppopts.texi: Document -pthread.
41750 2017-01-08 Martin Sebor <msebor@redhat.com>
41752 PR middle-end/77708
41753 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
41754 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
41755 New member functions.
41756 (format_directive): Used them.
41758 (pass_sprintf_length::handle_gimple_call): Same.
41759 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
41760 to avoid truncation for any argument.
41761 (extract_affine_mul): Same.
41762 * tree.c (get_file_function_name): Same.
41764 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41766 PR middle-end/77484
41767 * predict.def (PRED_INDIR_CALL): Set to 86.
41769 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41771 PR preprocessor/54124
41772 * doc/cppopts.texi: Reformat -d subtable to list the full name
41773 of the options. Add cross-reference to the docs for the general
41774 compiler -d options.
41775 * doc/invoke.texi (Developer Options): Add cross-reference to the
41776 preprocessor-specific -d option documentation.
41778 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41780 PR preprocessor/13498
41781 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
41782 redudant material, and reflect new command-line options.
41783 (System Headers): Likewise.
41785 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41787 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
41788 -isystem, and -idirafter. Copy-edit.
41789 * doc/cppopts.texi: Copy-edit. Remove contradiction about
41790 default for -ftrack-macro-expansion. Delete obsolete and
41791 badly-formatted implementation details about -fdebug-cpp output.
41792 * doc/cppwarnopts.texi: Copy-edit.
41794 2017-01-07 David Malcolm <dmalcolm@redhat.com>
41797 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41798 that the transition from a max line width >= 1<<10 to narrower
41799 lines works correctly.
41801 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
41803 * doc/options.texi (PerFunction): New.
41804 * opt-functions.awk (switch_flags): Map both Optimization and
41805 PerFunction to CL_OPTIMIZATION.
41806 * opth-gen.awk: Test for PerFunction flag along with
41808 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41809 it only when the latter is present. Skip those that don't in
41810 the hash function generator.
41811 * common.opt (fvar-tracking): Mark as PerFunction instead of
41813 (fvar-tracking-assignments): Likewise.
41814 (fvar-tracking-assignments-toggle): Likewise.
41815 (fvar-tracking-uninit): Likewise.
41817 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41819 PR translation/79018
41820 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41823 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41826 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41827 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41828 TARGET_LONG_JUMP_TABLE_OFFSETS.
41829 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41830 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41831 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41832 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41833 * config/m68k/m68k.md (tablejump expander): Likewise.
41834 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41835 TARGET_LONG_JUMP_TABLE_OFFSETS.
41836 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41837 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41839 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41840 David Holsgrove <david.holsgrove@xilinx.com>
41842 * common/config/microblaze/microblaze-common.c
41843 (TARGET_EXCEPT_UNWIND_INFO): Remove.
41844 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
41846 * config/microblaze/microblaze.c (microblaze_must_save_register)
41847 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41849 (microblaze_eh_return): New function.
41850 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41851 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41852 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41853 * config/microblaze/microblaze.md (eh_return): New pattern.
41855 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41857 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41858 GCC_DIAGNOSTIC_STRINGIFY): Define.
41860 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41862 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41864 * config/arm/arm.md (<mcrr>): New.
41866 * config/arm/arm.c (arm_arch5te): New.
41867 (arm_option_override): Set arm_arch5te.
41868 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41870 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41871 (arm_mcrr_qualifiers): ... this. New.
41872 (MRRC_QUALIFIERS): Define to...
41873 (arm_mrrc_qualifiers): ... this. New.
41874 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41876 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41877 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41878 (MRRCI, mrrc, MRRC): New.
41879 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41880 VUNSPEC_MRRC2): New.
41882 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41884 * config/arm/arm.md (<mcr>): New.
41886 * config/arm/arm.c (arm_coproc_builtin_available): Add
41887 support for mcr, mrc, mcr2 and mrc2.
41888 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41889 (arm_mcr_qualifiers): ... this. New.
41890 (MRC_QUALIFIERS): Define to ...
41891 (arm_mrc_qualifiers): ... this. New.
41892 (MCR_QUALIFIERS): Define to ...
41893 (arm_mcr_qualifiers): ... this. New.
41894 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41896 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41897 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41898 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41899 VUNSPEC_MRC2): New.
41901 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41903 * config/arm/arm.md (*ldc): New.
41907 * config/arm/arm.c (arm_coproc_builtin_available): Add
41908 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41909 (arm_coproc_ldc_stc_legitimate_address): New.
41910 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41911 'qualifier_const_pointer'.
41912 (LDC_QUALIFIERS): Define to...
41913 (arm_ldc_qualifiers): ... this. New.
41914 (STC_QUALIFIERS): Define to...
41915 (arm_stc_qualifiers): ... this. New.
41916 * config/arm/arm-protos.h
41917 (arm_coproc_ldc_stc_legitimate_address): New.
41918 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
41919 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
41920 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
41921 stc2, stcl, stc2l): New.
41922 * config/arm/constraints.md (Uz): New.
41923 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
41924 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
41925 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
41926 VUNSPEC_STC2L): New.
41928 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41930 * config/arm/arm.md (<cdp>): New.
41931 * config/arm/arm.c (neon_const_bounds): Rename this ...
41932 (arm_const_bounds): ... this.
41933 (arm_coproc_builtin_available): New.
41934 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
41935 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
41936 (CDP_QUALIFIERS): Define to...
41937 (arm_cdp_qualifiers): ... this. New.
41939 (arm_expand_builtin_args): Add case for 6 arguments.
41940 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
41941 (arm_const_bounds): ... this.
41942 (arm_coproc_builtin_available): New.
41943 * config/arm/arm_acle.h (__arm_cdp): New.
41945 * config/arm/arm_acle_builtins.def (cdp): New.
41947 * config/arm/iterators.md (CDPI,CDP,cdp): New.
41948 * config/arm/neon.md: Rename all 'neon_const_bounds' to
41949 'arm_const_bounds'.
41950 * config/arm/types.md (coproc): New.
41951 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
41952 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
41953 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
41954 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
41956 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41958 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
41959 (UBINOP_QUALIFIERS): New.
41961 (acle_builtin_data): New. Change comment.
41962 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
41963 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
41964 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
41965 arm_acle_builtins.def.
41966 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
41967 (arm_init_acle_builtins): New.
41968 (CRC32_BUILTIN): Remove.
41969 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
41970 crc32cb, crc32ch and crc32cw.
41971 (arm_init_crc32_builtins): Remove.
41972 (arm_init_builtins): Use arm_init_acle_builtins rather
41973 than arm_init_crc32_builtins.
41974 (arm_expand_acle_builtin): New.
41975 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
41976 * config/arm/arm_acle_builtins.def: New.
41978 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41980 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
41981 (arm_builtin_datum): ... this.
41982 (arm_init_neon_builtin): Rename to ...
41983 (arm_init_builtin): ... this. Add a new parameters PREFIX
41984 and USE_SIG_IN_NAME.
41985 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
41986 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
41987 'arm_builtin_datum'.
41988 (arm_init_vfp_builtins): Likewise.
41989 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
41990 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
41991 (arm_expand_neon_args): Rename to ...
41992 (arm_expand_builtin_args): ... this. Rename builtin_arg
41993 enum values and differentiate between ARG_BUILTIN_MEMORY
41994 and ARG_BUILTIN_NEON_MEMORY.
41995 (arm_expand_neon_builtin_1): Rename to ...
41996 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
41997 values, arm_expand_builtin_args and add bool parameter NEON.
41998 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
41999 (arm_expand_vfp_builtin): Likewise.
42000 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
42002 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42004 PR middle-end/77484
42005 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
42006 * predict.c (tree_estimate_probability_bb): Reverse direction of
42007 polymorphic call predictor.
42009 2017-01-06 David Malcolm <dmalcolm@redhat.com>
42011 * passes.c (execute_one_pass): Split out pass-skipping logic into...
42012 (determine_pass_name_match): ...this new function and...
42013 (should_skip_pass_p): ...this new function.
42015 2017-01-06 Nathan Sidwell <nathan@acm.org>
42017 * ipa-visibility.c (function_and_variable_visibility): Reformat
42018 comments and long lines. Remove extrneous if.
42019 * symtab.c (symtab_node::make_decl_local): Fix code format.
42020 (symtab_node::set_section_for_node): Fix comment typo.
42022 2017-01-06 Martin Liska <mliska@suse.cz>
42025 * lra-constraints.c: Rename invariant to lra_invariant.
42026 * predict.c (set_even_probabilities): Initialize e to NULL.
42028 2017-01-05 Martin Sebor <msebor@redhat.com>
42030 PR tree-optimization/78910
42031 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
42032 (format_integer): Correct off-by-one error in the handling
42033 of precision with negative numbers in signed conversions..
42035 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
42037 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
42039 2017-01-05 Jakub Jelinek <jakub@redhat.com>
42041 PR tree-optimization/71016
42042 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
42043 factor_out_conditional_conversion. Formatting fix.
42044 (factor_out_conditional_conversion): Add cond_stmt argument.
42045 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
42046 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
42049 2017-01-05 David Malcolm <dmalcolm@redhat.com>
42051 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
42052 read-rtl-function.o, and selftest-rtl.o.
42053 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
42054 (selftest::aarch64_test_loading_full_dump): New function.
42055 (selftest::aarch64_run_selftests): New function.
42056 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
42057 selftest::aarch64_run_selftests.
42058 * config/i386/i386.c
42059 (selftest::ix86_test_loading_dump_fragment_1): New function.
42060 (selftest::ix86_test_loading_call_insn): New function.
42061 (selftest::ix86_test_loading_full_dump): New function.
42062 (selftest::ix86_test_loading_unspec): New function.
42063 (selftest::ix86_run_selftests): Call the new functions.
42064 * emit-rtl.c (maybe_set_max_label_num): New function.
42065 * emit-rtl.h (maybe_set_max_label_num): New decl.
42066 * function.c (instantiate_decls): Guard call to
42067 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
42068 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
42070 * gensupport.c (gen_reader::gen_reader): Pass "false"
42071 for new "compact" param of rtx_reader.
42072 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
42073 rather than an empty string for NULL strings.
42074 * read-md.c: Potentially include config.h rather than bconfig.h.
42075 Wrap include of errors.h with #ifdef GENERATOR_FILE.
42076 (have_error): New global, copied from errors.c.
42077 (md_reader::read_name): Rename to...
42078 (md_reader::read_name_1): ...this, adding "out_loc" param,
42079 and converting "missing name or number" to returning false, rather
42081 (md_reader::read_name): Reimplement in terms of read_name_1.
42082 (md_reader::read_name_or_nil): New function.
42083 (md_reader::read_string): Handle "(nil)" by returning NULL.
42084 (md_reader::md_reader): Add new param "compact".
42085 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
42086 (md_reader::read_file): New method.
42087 * read-md.h (md_reader::md_reader): Add new param "compact".
42088 (md_reader::read_file): New method.
42089 (md_reader::is_compact): New accessor.
42090 (md_reader::read_name): Convert return type from void to file_location.
42091 (md_reader::read_name_or_nil): New decl.
42092 (md_reader::read_name_1): New decl.
42093 (md_reader::m_compact): New field.
42094 (noop_reader::noop_reader): Pass "false" for new "compact" param
42096 (rtx_reader::rtx_reader): Add new "compact" param.
42097 (rtx_reader::read_rtx_operand): Make virtual and convert return
42098 type from void to rtx.
42099 (rtx_reader::read_until): New decl.
42100 (rtx_reader::handle_any_trailing_information): New virtual function.
42101 (rtx_reader::postprocess): New virtual function.
42102 (rtx_reader::finalize_string): New virtual function.
42103 (rtx_reader::m_in_call_function_usage): New field.
42104 (rtx_reader::m_reuse_rtx_by_id): New field.
42105 * read-rtl-function.c: New file.
42106 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
42107 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
42108 (selftest::verify_three_block_rtl_cfg): New decl.
42109 * read-rtl-function.h: New file.
42110 * read-rtl.c: Potentially include config.h rather than bconfig.h.
42111 For host, include function.h, memmodel.h, and emit-rtl.h.
42112 (one_time_initialization): New function.
42113 (struct compact_insn_name): New struct.
42114 (compact_insn_names): New array.
42115 (find_code): Handle insn codes in compact dumps.
42116 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
42117 (bind_subst_iter_and_attr): Likewise.
42118 (add_condition_to_string): Likewise.
42119 (add_condition_to_rtx): Likewise.
42120 (apply_attribute_uses): Likewise.
42121 (add_current_iterators): Likewise.
42122 (apply_iterators): Likewise.
42123 (initialize_iterators): Guard usage of apply_subst_iterator with
42124 #ifdef GENERATOR_FILE.
42125 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
42126 (md_reader::read_mapping): Likewise.
42127 (add_define_attr_for_define_subst): Likewise.
42128 (add_define_subst_attr): Likewise.
42129 (read_subst_mapping): Likewise.
42130 (check_code_iterator): Likewise.
42131 (rtx_reader::read_rtx): Likewise. Move one-time initialization
42133 (one_time_initialization): New function.
42134 (rtx_reader::read_until): New method.
42135 (read_flags): New function.
42136 (parse_reg_note_name): New function.
42137 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
42138 Handle reuse_rtx ids.
42139 Wrap iterator lookup within #ifdef GENERATOR_FILE.
42140 Add parsing support for RTL dumps, mirroring the special-cases in
42141 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
42142 values, and calling handle_any_trailing_information.
42143 (rtx_reader::read_rtx_operand): Convert return type from void
42144 to rtx, returning return_rtx. Handle case 'e'. Call
42145 finalize_string on XSTR and XTMPL fields.
42146 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
42147 "(nil)" values were omitted. Call the postprocess vfunc on the
42149 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
42150 class ctor. Initialize m_in_call_function_usage. Call
42151 one_time_initialization.
42152 * rtl-tests.c (selftest::test_uncond_jump): Call
42153 set_new_first_and_last_insn.
42154 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
42155 * selftest-rtl.c: New file.
42156 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
42157 (selftest::get_insn_by_uid): New decl.
42158 * selftest-run-tests.c (selftest::run_tests): Call
42159 read_rtl_function_c_tests.
42160 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
42161 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
42164 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
42166 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
42167 operands in a special way. Assert that pos+len <= mode precision.
42169 2017-01-05 Jakub Jelinek <jakub@redhat.com>
42171 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
42172 3 argument Alias with unlimited for the negative form.
42173 (fno-vect-cost-model): Removed.
42175 2017-01-05 Martin Liska <mliska@suse.cz>
42177 * hsa-gen.c (gen_hsa_divmod): New function.
42178 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
42180 2017-01-05 Martin Liska <mliska@suse.cz>
42183 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
42186 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42188 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
42189 small constant length operands.
42191 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42193 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
42194 between loop iterations.
42196 2017-01-05 Martin Liska <mliska@suse.cz>
42199 * gimplify.c (gimplify_decl_expr): Compare to
42200 asan_poisoned_variables instread of checking flags.
42201 (gimplify_target_expr): Likewise.
42202 (gimplify_expr): Likewise.
42203 (gimplify_function_tree): Conditionally initialize
42204 asan_poisoned_variables.
42206 2017-01-04 Jeff Law <law@redhat.com>
42208 PR tree-optimizatin/78812
42209 * rtl.h (contains_mem_rtx_p): Prototype.
42210 * ifcvt.c (containts_mem_rtx_p): Move from here to...
42211 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
42212 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
42213 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
42214 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
42216 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42218 * input.c (assert_char_at_range): Default-initialize actual_range.
42220 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42222 * df-scan.c (df_ref_create_structure): Make regno unsigned,
42223 to match the caller.
42225 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42227 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
42228 insns after final jump in test to emit dummy move.
42230 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42232 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
42233 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
42235 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42237 * multiple_target.c (create_dispatcher_calls): Init e_next.
42238 * tree-ssa-loop-split.c (split_loop): Init border.
42239 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
42242 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
42247 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
42248 (altivec_register_operand): Do not return true if the operand
42249 contains a SUBREG mixing SImode and SFmode.
42250 (vsx_register_operand): Likewise.
42251 (vsx_reg_sfsubreg_ok): New predicate.
42252 (vfloat_operand): Do not return true if the operand contains a
42253 SUBREG mixing SImode and SFmode.
42254 (vint_operand): Likewise.
42255 (vlogical_operand): Likewise.
42256 (gpc_reg_operand): Likewise.
42257 (int_reg_operand): Likewise.
42258 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
42259 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
42260 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
42262 (rs6000_emit_move_si_sf_subreg): New helper function.
42263 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
42264 fixup SUBREGs involving SImode and SFmode.
42265 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
42266 numbers for the new peephole2 optimization.
42267 (peephole2 for SFmode unions): New peephole2 to optimize cases in
42268 the GLIBC math library that do AND/IOR/XOR operations on single
42269 precision floating point.
42270 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
42271 target macros to say whether we need to avoid SUBREGs mixing
42273 (TARGET_ALLOW_SF_SUBREG): Likewise.
42274 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
42275 (UNSPEC_SI_FROM_SF): Likewise.
42276 (iorxor): Change spacing.
42277 (and_ior_xor): New iterator for AND, IOR, and XOR.
42278 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
42279 (movdi_from_sf_zero_ext): Likewise.
42280 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
42281 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
42282 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
42283 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
42284 (fms<mode>4): Likewise.
42285 (fnma<mode>4): Likewise.
42286 (fnms<mode>4): Likewise.
42287 (nfma<mode>4): Likewise.
42288 (nfms<mode>4): Likewise.
42290 2017-01-04 Marek Polacek <polacek@redhat.com>
42293 * doc/invoke.texi: Document -Wpointer-compare.
42295 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42297 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
42300 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
42301 descriptions for -gdwarf-5 and emit them as uleb128 instead of
42304 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
42307 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
42308 documentation of the powerpc_popcntb_ok attribute.
42309 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
42310 code to issue warning messages if a requested CPU configuration is
42311 not supported by the binary (assembler and loader) toolchain.
42312 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
42313 made to define a built-in function that has been disabled.
42314 (paired_init_builtins): Add assertion to prevent ICE if attempt is
42315 made to define a built-in function that has been disabled.
42316 (altivec_init_builtins): Add comment explaining why definition
42317 of the DST built-in functions is not preceded by an assertion
42318 check. Add assertions to prevent ICE if attempts are made to
42319 define an altivec predicate or an abs* built-in function that has
42321 (htm_init_builtins): Add comment explaining why definition of the
42322 htm built-in functions is not preceded by an assertion check.
42324 2017-01-04 Jeff Law <law@redhat.com>
42326 PR tree-optimizatin/67955
42327 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
42328 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
42329 the points-to solution does not include pt_null. Use DECL_PT_UID
42332 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
42334 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
42335 Use gen_int_mode instead of gen_lopwart for const_int operands.
42337 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42339 PR tree-optimization/71563
42340 * match.pd: Simplify X << Y into X if Y is known to be 0 or
42341 out of range value - has low bits known to be zero.
42343 2017-01-04 Alan Modra <amodra@gmail.com>
42345 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
42346 * configure: Regenerate.
42347 * config.in: Regenerate.
42349 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42352 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
42353 a substring of the message, but strcmp with the whole message. Ifdef
42354 ENABLE_NLS, translate the message first using dgettext.
42356 2017-01-03 Jeff Law <law@redhat.com>
42358 PR tree-optimizatin/78856
42359 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
42360 (mark_threaded_blocks): Remove code to truncate thread paths that
42361 cross multiple loop headers. Instead invalidate the cached loop
42362 iteration information and handle case of a thread path walking
42363 into an irreducible region.
42365 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
42368 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
42369 assertions. Add support for doing the signbit if the IEEE 128-bit
42370 floating point value is in a GPR.
42371 * config/rs6000/rs6000.md (Fsignbit): Delete.
42372 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
42373 Update the length attribute if the value is in a GPR.
42374 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
42375 the sign or zero extension instruction, since the value is always 0/1.
42376 (signbit<mode>2_dm2): Delete using <Fsignbit>.
42379 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
42380 extracting SImode to a GPR register so that we can generate a
42381 store, limit the vector to be in a traditional Altivec register
42382 for the vextuwrx instruction.
42384 2017-01-03 Ian Lance Taylor <iant@google.com>
42386 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
42388 2017-01-03 Martin Sebor <msebor@redhat.com>
42390 PR tree-optimization/78696
42391 * gimple-ssa-sprintf.c (format_floating): Correct handling of
42392 precision. Use MPFR for %f for greater fidelity. Correct handling
42394 (pass_sprintf_length::compute_format_length): Set width and precision
42395 specified by asrerisk to void_node for vararg functions.
42396 (try_substitute_return_value): Adjust dump output.
42398 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
42400 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
42402 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
42404 * doc/invoke.texi (SPARC options): Document -mlra as the default.
42405 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
42406 -mlra/-mno-lra was passed to the compiler.
42408 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
42410 PR rtl-optimization/65618
42411 * emit-rtl.c (try_split): Move initialization of "before" and
42412 "after" to just before the call to emit_insn_after_setloc.
42414 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
42416 * doc/md.texi (Standard Names): Remove reference to Java frontend.
42418 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
42420 * dwarf2out.c (gen_enumeration_type_die): When
42421 -gno-strict-dwarf, add a DW_AT_encoding attribute.
42423 2017-01-03 Jakub Jelinek <jakub@redhat.com>
42425 PR tree-optimization/78965
42426 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
42427 Change first argument from const call_info & to call_info &. For %n
42428 set info.nowrite to false.
42430 PR middle-end/78901
42431 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
42432 possibly throwing calls.
42434 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
42435 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
42436 and fns handling, rather than in a separate case SSA_NAME.
42438 2017-01-02 Jeff Law <law@redhat.com>
42440 * config/darwin-driver.c (darwin_driver_init): Const-correctness
42441 fixes for first_period and second_period variables.
42443 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
42446 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
42447 (*insvqi_1): New insn pattern.
42448 (*insvqi_1_mem_rex64): Ditto.
42449 (*insvqi_2): Ditto.
42450 (*insvqi_3): Rename from *insvqi.
42452 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
42454 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
42456 * doc/cfg.texi (Edges): Remove reference to Java.
42457 (Maintaining the CFG): Ditto.
42459 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42461 PR middle-end/77674
42462 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
42463 transparent aliases.
42465 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42467 PR middle-end/77484
42468 * predict.def (PRED_CALL): Update hitrate.
42469 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
42470 * predict.c (tree_estimate_probability_bb): Split CALL predictor
42471 into direct/indirect/polymorphic variants.
42473 2017-01-01 Jakub Jelinek <jakub@redhat.com>
42475 Update copyright years.
42477 * gcc.c (process_command): Update copyright notice dates.
42478 * gcov-dump.c (print_version): Ditto.
42479 * gcov.c (print_version): Ditto.
42480 * gcov-tool.c (print_version): Ditto.
42481 * gengtype.c (create_file): Ditto.
42482 * doc/cpp.texi: Bump @copying's copyright year.
42483 * doc/cppinternals.texi: Ditto.
42484 * doc/gcc.texi: Ditto.
42485 * doc/gccint.texi: Ditto.
42486 * doc/gcov.texi: Ditto.
42487 * doc/install.texi: Ditto.
42488 * doc/invoke.texi: Ditto.
42490 Copyright (C) 2017 Free Software Foundation, Inc.
42492 Copying and distribution of this file, with or without modification,
42493 are permitted in any medium without royalty provided the copyright
42494 notice and this notice are preserved.