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-slp.c (vect_get_and_check_slp_defs): Reject
6 constant and extern definitions for variable-length vectors.
7 (vect_get_constant_vectors): Note that the number of units
8 is known to be constant.
10 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
11 Alan Hayward <alan.hayward@arm.com>
12 David Sherwood <david.sherwood@arm.com>
14 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
15 of units as polynomial. Choose between WIDE and NARROW based
18 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
19 Alan Hayward <alan.hayward@arm.com>
20 David Sherwood <david.sherwood@arm.com>
22 * tree-vect-stmts.c (simd_clone_subparts): New function.
23 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
25 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
26 Alan Hayward <alan.hayward@arm.com>
27 David Sherwood <david.sherwood@arm.com>
29 * tree-vect-stmts.c (vectorizable_call): Treat the number of
30 vectors as polynomial. Use build_index_vector for
33 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
34 Alan Hayward <alan.hayward@arm.com>
35 David Sherwood <david.sherwood@arm.com>
37 * tree-vect-stmts.c (get_load_store_type): Treat the number of
38 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
39 for variable-length vectors.
40 (vectorizable_mask_load_store): Treat the number of units as
41 polynomial, asserting that it is constant if the condition has
42 already been enforced.
43 (vectorizable_store, vectorizable_load): Likewise.
45 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
46 Alan Hayward <alan.hayward@arm.com>
47 David Sherwood <david.sherwood@arm.com>
49 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
50 of units as polynomial. Punt if we can't tell at compile time
51 which vector contains the final result.
53 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
54 Alan Hayward <alan.hayward@arm.com>
55 David Sherwood <david.sherwood@arm.com>
57 * tree-vect-loop.c (vectorizable_induction): Treat the number
58 of units as polynomial. Punt on SLP inductions. Use an integer
59 VEC_SERIES_EXPR for variable-length integer reductions. Use a
60 cast of such a series for variable-length floating-point
63 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
64 Alan Hayward <alan.hayward@arm.com>
65 David Sherwood <david.sherwood@arm.com>
67 * tree.h (build_index_vector): Declare.
68 * tree.c (build_index_vector): New function.
69 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
70 of units as polynomial, forcibly converting it to a constant if
71 vectorizable_reduction has already enforced the condition.
72 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
73 to create a {1,2,3,...} vector.
74 (vectorizable_reduction): Treat the number of units as polynomial.
75 Choose vectype_in based on the largest scalar element size rather
76 than the smallest number of units. Enforce the restrictions
79 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
80 Alan Hayward <alan.hayward@arm.com>
81 David Sherwood <david.sherwood@arm.com>
83 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
84 number of units as polynomial.
86 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
87 Alan Hayward <alan.hayward@arm.com>
88 David Sherwood <david.sherwood@arm.com>
90 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
91 * target.def (autovectorize_vector_sizes): Return the vector sizes
92 by pointer, using vector_sizes rather than a bitmask.
93 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
94 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
95 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
97 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
98 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
99 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
100 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
101 * omp-general.c (omp_max_vf): Likewise.
102 * omp-low.c (omp_clause_aligned_alignment): Likewise.
103 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
104 * tree-vect-loop.c (vect_analyze_loop): Likewise.
105 * tree-vect-slp.c (vect_slp_bb): Likewise.
106 * doc/tm.texi: Regenerate.
107 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
109 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
110 the vector size as a poly_uint64 rather than an unsigned int.
111 (current_vector_size): Change from an unsigned int to a poly_uint64.
112 (get_vectype_for_scalar_type): Update accordingly.
113 * tree.h (build_truth_vector_type): Take the size and number of
114 units as a poly_uint64 rather than an unsigned int.
115 (build_vector_type): Add a temporary overload that takes
116 the number of units as a poly_uint64 rather than an unsigned int.
117 * tree.c (make_vector_type): Likewise.
118 (build_truth_vector_type): Take the number of units as a poly_uint64
119 rather than an unsigned int.
121 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
122 Alan Hayward <alan.hayward@arm.com>
123 David Sherwood <david.sherwood@arm.com>
125 * target.def (get_mask_mode): Take the number of units and length
126 as poly_uint64s rather than unsigned ints.
127 * targhooks.h (default_get_mask_mode): Update accordingly.
128 * targhooks.c (default_get_mask_mode): Likewise.
129 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
130 * doc/tm.texi: Regenerate.
132 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
133 Alan Hayward <alan.hayward@arm.com>
134 David Sherwood <david.sherwood@arm.com>
136 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
137 * omp-general.c (omp_max_vf): Likewise.
138 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
139 (expand_omp_simd): Handle polynomial safelen.
140 * omp-low.c (omplow_simd_context): Add a default constructor.
141 (omplow_simd_context::max_vf): Change from int to poly_uint64.
142 (lower_rec_simd_input_clauses): Update accordingly.
143 (lower_rec_input_clauses): Likewise.
145 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
146 Alan Hayward <alan.hayward@arm.com>
147 David Sherwood <david.sherwood@arm.com>
149 * tree-vectorizer.h (vect_nunits_for_cost): New function.
150 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
151 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
152 (vect_analyze_slp_cost): Likewise.
153 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
154 (vect_model_load_cost): Likewise.
156 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
157 Alan Hayward <alan.hayward@arm.com>
158 David Sherwood <david.sherwood@arm.com>
160 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
161 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
162 from an unsigned int * to a poly_uint64_pod *.
163 (calculate_unrolling_factor): New function.
164 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
166 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
167 Alan Hayward <alan.hayward@arm.com>
168 David Sherwood <david.sherwood@arm.com>
170 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
171 from an unsigned int to a poly_uint64.
172 (_loop_vec_info::slp_unrolling_factor): Likewise.
173 (_loop_vec_info::vectorization_factor): Change from an int
175 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
176 (vect_get_num_vectors): New function.
177 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
178 (vect_get_num_copies): Use vect_get_num_vectors.
179 (vect_analyze_data_ref_dependences): Change max_vf from an int *
180 to an unsigned int *.
181 (vect_analyze_data_refs): Change min_vf from an int * to a
183 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
184 than an unsigned HOST_WIDE_INT.
185 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
186 (vect_analyze_data_ref_dependence): Change max_vf from an int *
187 to an unsigned int *.
188 (vect_analyze_data_ref_dependences): Likewise.
189 (vect_compute_data_ref_alignment): Handle polynomial vf.
190 (vect_enhance_data_refs_alignment): Likewise.
191 (vect_prune_runtime_alias_test_list): Likewise.
192 (vect_shift_permute_load_chain): Likewise.
193 (vect_supportable_dr_alignment): Likewise.
194 (dependence_distance_ge_vf): Take the vectorization factor as a
195 poly_uint64 rather than an unsigned HOST_WIDE_INT.
196 (vect_analyze_data_refs): Change min_vf from an int * to a
198 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
199 vfm1 as a poly_uint64 rather than an int. Make the same change
200 for the returned bound_scalar.
201 (vect_gen_vector_loop_niters): Handle polynomial vf.
202 (vect_do_peeling): Likewise. Update call to
203 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
204 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
206 * tree-vect-loop.c (vect_determine_vectorization_factor)
207 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
208 (vect_get_known_peeling_cost): Likewise.
209 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
210 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
211 (vect_transform_loop): Likewise. Use the lowest possible VF when
212 updating the upper bounds of the loop.
213 (vect_min_worthwhile_factor): Make static. Return an unsigned int
215 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
216 polynomial unroll factors.
217 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
218 (vect_make_slp_decision): Likewise.
219 (vect_supported_load_permutation_p): Likewise, and polynomial
221 (vect_analyze_slp_cost): Handle polynomial vf.
222 (vect_slp_analyze_node_operations): Likewise.
223 (vect_slp_analyze_bb_1): Likewise.
224 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
225 than an unsigned HOST_WIDE_INT.
226 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
227 (vectorizable_load): Handle polynomial vf.
228 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
230 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
232 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
233 Alan Hayward <alan.hayward@arm.com>
234 David Sherwood <david.sherwood@arm.com>
236 * match.pd: Handle bit operations involving three constants
237 and try to fold one pair.
239 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
241 * tree-vect-loop-manip.c: Include gimple-fold.h.
242 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
243 niters_maybe_zero parameters. Handle other cases besides a step of 1.
244 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
245 Add a path that uses a step of VF instead of 1, but disable it
247 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
248 and niters_no_overflow parameters. Update calls to
249 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
250 Create a new SSA name if the latter choses to use a ste other
251 than zero, and return it via niters_vector_mult_vf_var.
252 * tree-vect-loop.c (vect_transform_loop): Update calls to
253 vect_do_peeling, vect_gen_vector_loop_niters and
254 slpeel_make_loop_iterate_ntimes.
255 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
256 (vect_gen_vector_loop_niters): Update declarations after above changes.
258 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
260 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
261 128-bit round to integer instructions.
262 (ceil<mode>2): Likewise.
263 (btrunc<mode>2): Likewise.
264 (round<mode>2): Likewise.
266 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
268 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
269 unaligned VSX load/store on P8/P9.
270 (expand_block_clear): Allow the use of unaligned VSX
273 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
275 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
277 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
278 swap associated with both a load and a store.
280 2018-01-02 Andrew Waterman <andrew@sifive.com>
282 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
283 * config/riscv/riscv.md (clear_cache): Use it.
285 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
287 * web.c: Remove out-of-date comment.
289 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
291 * expr.c (fixup_args_size_notes): Check that any existing
292 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
293 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
294 (emit_single_push_insn): ...here.
296 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
298 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
299 (const_vector_encoded_nelts): New function.
300 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
301 (const_vector_int_elt, const_vector_elt): Declare.
302 * emit-rtl.c (const_vector_int_elt_1): New function.
303 (const_vector_elt): Likewise.
304 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
307 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
309 * expr.c: Include rtx-vector-builder.h.
310 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
311 directly on the tree encoding.
312 (const_vector_from_tree): Likewise.
313 * optabs.c: Include rtx-vector-builder.h.
314 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
315 sequence of "u" values.
316 * vec-perm-indices.c: Include rtx-vector-builder.h.
317 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
318 directly on the vec_perm_indices encoding.
320 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
322 * doc/rtl.texi (const_vector): Describe new encoding scheme.
323 * Makefile.in (OBJS): Add rtx-vector-builder.o.
324 * rtx-vector-builder.h: New file.
325 * rtx-vector-builder.c: Likewise.
326 * rtl.h (rtx_def::u2): Add a const_vector field.
327 (CONST_VECTOR_NPATTERNS): New macro.
328 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
329 (CONST_VECTOR_DUPLICATE_P): Likewise.
330 (CONST_VECTOR_STEPPED_P): Likewise.
331 (CONST_VECTOR_ENCODED_ELT): Likewise.
332 (const_vec_duplicate_p): Check for a duplicated vector encoding.
333 (unwrap_const_vec_duplicate): Likewise.
334 (const_vec_series_p): Check for a non-duplicated vector encoding.
335 Say that the function only returns true for integer vectors.
336 * emit-rtl.c: Include rtx-vector-builder.h.
337 (gen_const_vec_duplicate_1): Delete.
338 (gen_const_vector): Call gen_const_vec_duplicate instead of
339 gen_const_vec_duplicate_1.
340 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
341 (gen_const_vec_duplicate): Use rtx_vector_builder.
342 (gen_const_vec_series): Likewise.
343 (gen_rtx_CONST_VECTOR): Likewise.
344 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
345 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
346 Build a new vector rather than modifying a CONST_VECTOR in-place.
347 (handle_special_swappables): Update call accordingly.
348 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
349 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
350 Build a new vector rather than modifying a CONST_VECTOR in-place.
351 (handle_special_swappables): Update call accordingly.
353 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
355 * simplify-rtx.c (simplify_const_binary_operation): Use
356 CONST_VECTOR_ELT instead of XVECEXP.
358 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
360 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
361 the selector elements to be different from the data elements
362 if the selector is a VECTOR_CST.
363 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
364 ssizetype for the selector.
366 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
368 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
369 before testing each element individually.
370 * tree-vect-generic.c (lower_vec_perm): Likewise.
372 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
374 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
375 * selftest-run-tests.c (selftest::run_tests): Call it.
376 * vector-builder.h (vector_builder::operator ==): New function.
377 (vector_builder::operator !=): Likewise.
378 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
379 (vec_perm_indices::all_from_input_p): New function.
380 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
381 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
382 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
383 instead of reading the VECTOR_CST directly. Detect whether both
384 vector inputs are the same before constructing the vec_perm_indices,
385 and update the number of inputs argument accordingly. Use the
386 utility functions added above. Only construct sel2 if we need to.
388 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
390 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
391 the broadcast of the low byte.
392 (expand_mult_highpart): Use an explicit encoding for the permutes.
393 * optabs-query.c (can_mult_highpart_p): Likewise.
394 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
395 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
396 (vectorizable_bswap): Likewise.
397 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
398 explicit encoding for the power-of-2 permutes.
399 (vect_permute_store_chain): Likewise.
400 (vect_grouped_load_supported): Likewise.
401 (vect_permute_load_chain): Likewise.
403 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
405 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
406 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
407 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
408 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
409 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
410 (vect_gen_perm_mask_any): Likewise.
412 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
414 * int-vector-builder.h: New file.
415 * vec-perm-indices.h: Include int-vector-builder.h.
416 (vec_perm_indices): Redefine as an int_vector_builder.
417 (auto_vec_perm_indices): Delete.
418 (vec_perm_builder): Redefine as a stand-alone class.
419 (vec_perm_indices::vec_perm_indices): New function.
420 (vec_perm_indices::clamp): Likewise.
421 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
422 (vec_perm_indices::new_vector): New function.
423 (vec_perm_indices::new_expanded_vector): Update for new
424 vec_perm_indices class.
425 (vec_perm_indices::rotate_inputs): New function.
426 (vec_perm_indices::all_in_range_p): Operate directly on the
427 encoded form, without computing elided elements.
428 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
429 encoding. Update for new vec_perm_indices class.
430 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
431 the given vec_perm_builder.
432 (expand_vec_perm_var): Update vec_perm_builder constructor.
433 (expand_mult_highpart): Use vec_perm_builder instead of
434 auto_vec_perm_indices.
435 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
436 vec_perm_indices instead of auto_vec_perm_indices. Use a single
437 or double series encoding as appropriate.
438 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
439 vec_perm_indices instead of auto_vec_perm_indices.
440 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
441 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
442 (vect_permute_store_chain): Likewise.
443 (vect_grouped_load_supported): Likewise.
444 (vect_permute_load_chain): Likewise.
445 (vect_shift_permute_load_chain): Likewise.
446 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
447 (vect_transform_slp_perm_load): Likewise.
448 (vect_schedule_slp_instance): Likewise.
449 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
450 (vectorizable_mask_load_store): Likewise.
451 (vectorizable_bswap): Likewise.
452 (vectorizable_store): Likewise.
453 (vectorizable_load): Likewise.
454 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
455 vec_perm_indices instead of auto_vec_perm_indices. Use
456 tree_to_vec_perm_builder to read the vector from a tree.
457 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
458 vec_perm_builder instead of a vec_perm_indices.
459 (have_whole_vector_shift): Use vec_perm_builder and
460 vec_perm_indices instead of auto_vec_perm_indices. Leave the
461 truncation to calc_vec_perm_mask_for_shift.
462 (vect_create_epilog_for_reduction): Likewise.
463 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
464 from auto_vec_perm_indices to vec_perm_indices.
465 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
466 instead of changing individual elements.
467 (aarch64_vectorize_vec_perm_const): Use new_vector to install
468 the vector in d.perm.
469 * config/arm/arm.c (expand_vec_perm_d::perm): Change
470 from auto_vec_perm_indices to vec_perm_indices.
471 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
472 instead of changing individual elements.
473 (arm_vectorize_vec_perm_const): Use new_vector to install
474 the vector in d.perm.
475 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
476 Update vec_perm_builder constructor.
477 (rs6000_expand_interleave): Likewise.
478 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
479 (rs6000_expand_interleave): Likewise.
481 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
483 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
484 to qimode could truncate the indices.
485 * optabs.c (expand_vec_perm_var): Likewise.
487 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
489 * Makefile.in (OBJS): Add vec-perm-indices.o.
490 * vec-perm-indices.h: New file.
491 * vec-perm-indices.c: Likewise.
492 * target.h (vec_perm_indices): Replace with a forward class
494 (auto_vec_perm_indices): Move to vec-perm-indices.h.
495 * optabs.h: Include vec-perm-indices.h.
496 (expand_vec_perm): Delete.
497 (selector_fits_mode_p, expand_vec_perm_var): Declare.
498 (expand_vec_perm_const): Declare.
499 * target.def (vec_perm_const_ok): Replace with...
500 (vec_perm_const): ...this new hook.
501 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
502 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
503 * doc/tm.texi: Regenerate.
504 * optabs.def (vec_perm_const): Delete.
505 * doc/md.texi (vec_perm_const): Likewise.
506 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
507 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
508 expand_vec_perm for constant permutation vectors. Assert that
509 the mode of variable permutation vectors is the integer equivalent
510 of the mode that is being permuted.
511 * optabs-query.h (selector_fits_mode_p): Declare.
512 * optabs-query.c: Include vec-perm-indices.h.
513 (selector_fits_mode_p): New function.
514 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
515 is defined, instead of checking whether the vec_perm_const_optab
516 exists. Use targetm.vectorize.vec_perm_const instead of
517 targetm.vectorize.vec_perm_const_ok. Check whether the indices
518 fit in the vector mode before using a variable permute.
519 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
520 vec_perm_indices instead of an rtx.
521 (expand_vec_perm): Replace with...
522 (expand_vec_perm_const): ...this new function. Take the selector
523 as a vec_perm_indices rather than an rtx. Also take the mode of
524 the selector. Update call to shift_amt_for_vec_perm_mask.
525 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
526 Use vec_perm_indices::new_expanded_vector to expand the original
527 selector into bytes. Check whether the indices fit in the vector
528 mode before using a variable permute.
529 (expand_vec_perm_var): Make global.
530 (expand_mult_highpart): Use expand_vec_perm_const.
531 * fold-const.c: Includes vec-perm-indices.h.
532 * tree-ssa-forwprop.c: Likewise.
533 * tree-vect-data-refs.c: Likewise.
534 * tree-vect-generic.c: Likewise.
535 * tree-vect-loop.c: Likewise.
536 * tree-vect-slp.c: Likewise.
537 * tree-vect-stmts.c: Likewise.
538 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
540 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
541 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
542 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
543 (aarch64_vectorize_vec_perm_const): ...this new function.
544 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
545 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
546 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
547 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
548 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
549 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
550 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
552 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
553 check for NEON modes.
554 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
555 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
556 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
557 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
559 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
560 the old VEC_PERM_CONST conditions.
561 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
562 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
563 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
564 (ia64_vectorize_vec_perm_const_ok): Merge into...
565 (ia64_vectorize_vec_perm_const): ...this new function.
566 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
567 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
568 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
569 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
570 * config/mips/mips.c (mips_expand_vec_perm_const)
571 (mips_vectorize_vec_perm_const_ok): Merge into...
572 (mips_vectorize_vec_perm_const): ...this new function.
573 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
574 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
575 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
576 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
577 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
578 (rs6000_expand_vec_perm_const): Delete.
579 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
581 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
582 (altivec_expand_vec_perm_const_le): Take each operand individually.
583 Operate on constant selectors rather than rtxes.
584 (altivec_expand_vec_perm_const): Likewise. Update call to
585 altivec_expand_vec_perm_const_le.
586 (rs6000_expand_vec_perm_const): Delete.
587 (rs6000_vectorize_vec_perm_const_ok): Delete.
588 (rs6000_vectorize_vec_perm_const): New function.
589 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
590 an element count and rtx array.
591 (rs6000_expand_extract_even): Update call accordingly.
592 (rs6000_expand_interleave): Likewise.
593 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
594 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
595 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
596 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
597 (rs6000_expand_vec_perm_const): Delete.
598 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
599 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
600 (altivec_expand_vec_perm_const_le): Take each operand individually.
601 Operate on constant selectors rather than rtxes.
602 (altivec_expand_vec_perm_const): Likewise. Update call to
603 altivec_expand_vec_perm_const_le.
604 (rs6000_expand_vec_perm_const): Delete.
605 (rs6000_vectorize_vec_perm_const_ok): Delete.
606 (rs6000_vectorize_vec_perm_const): New function. Remove stray
607 reference to the SPE evmerge intructions.
608 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
609 an element count and rtx array.
610 (rs6000_expand_extract_even): Update call accordingly.
611 (rs6000_expand_interleave): Likewise.
612 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
613 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
615 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
617 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
619 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
620 vector mode and that that mode matches the mode of the data
622 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
623 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
624 directly using expand_vec_perm_1 when forcing selectors into
626 (expand_vec_perm_var): New function, split out from expand_vec_perm.
628 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
630 * optabs-query.h (can_vec_perm_p): Delete.
631 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
632 * optabs-query.c (can_vec_perm_p): Split into...
633 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
634 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
635 particular selector is valid.
636 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
637 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
638 (vect_grouped_load_supported): Likewise.
639 (vect_shift_permute_load_chain): Likewise.
640 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
641 (vect_transform_slp_perm_load): Likewise.
642 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
643 (vectorizable_bswap): Likewise.
644 (vect_gen_perm_mask_checked): Likewise.
645 * fold-const.c (fold_ternary_loc): Likewise. Don't take
646 implementations of variable permutation vectors into account
647 when deciding which selector to use.
648 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
649 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
650 with a false third argument.
651 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
652 to test whether the constant selector is valid and can_vec_perm_var_p
653 to test whether a variable selector is valid.
655 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
657 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
658 * optabs-query.c (can_vec_perm_p): Likewise.
659 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
660 instead of vec_perm_indices.
661 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
662 (vect_gen_perm_mask_checked): Likewise,
663 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
664 (vect_gen_perm_mask_checked): Likewise,
666 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
668 * optabs-query.h (qimode_for_vec_perm): Declare.
669 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
670 (qimode_for_vec_perm): ...this new function.
671 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
673 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
675 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
676 does not have a conditional at the top.
678 2018-01-02 Richard Biener <rguenther@suse.de>
680 * ipa-inline.c (big_speedup_p): Fix expression.
682 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
685 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
688 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
692 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
693 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
694 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
695 cond_taken_branch_cost 3->4.
697 2018-01-01 Jakub Jelinek <jakub@redhat.com>
699 PR tree-optimization/83581
700 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
701 TODO_cleanup_cfg if any changes have been made.
704 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
705 convert_modes if target mode has the right side, but different mode
709 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
710 last argument when extracting from CONCAT. If either from_real or
711 from_imag is NULL, use expansion through memory. If result is not
712 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
713 the parts directly to inner mode, if even that fails, use expansion
717 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
718 check for bswap in mode rather than HImode and use that in expand_unop
721 2017-12-31 Jakub Jelinek <jakub@redhat.com>
724 * config/i386/i386.c (ix86_attribute_table): Remove excess
725 initializer for "shared" attribute.
727 2017-12-30 Tom de Vries <tom@codesourcery.com>
730 * omp-expand.c (expand_omp_target): If in_lto_p, mark offload_funcs with
732 * lto-streamer-out.c (prune_offload_funcs): New function. Remove
733 offload_funcs entries that no longer have a corresponding cgraph_node.
734 Mark the remaining ones as DECL_PRESERVE_P.
735 (output_lto): Call prune_offload_funcs.
737 2017-12-30 Jakub Jelinek <jakub@redhat.com>
739 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
740 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>,
741 vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
742 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
743 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
744 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1, vpdpbusd_<mode>,
745 vpdpbusd_<mode>_mask, vpdpbusd_<mode>_maskz, vpdpbusd_<mode>_maskz_1,
746 vpdpbusds_<mode>, vpdpbusds_<mode>_mask, vpdpbusds_<mode>_maskz,
747 vpdpbusds_<mode>_maskz_1, vpdpwssd_<mode>, vpdpwssd_<mode>_mask,
748 vpdpwssd_<mode>_maskz, vpdpwssd_<mode>_maskz_1, vpdpwssds_<mode>,
749 vpdpwssds_<mode>_mask, vpdpwssds_<mode>_maskz,
750 vpdpwssds_<mode>_maskz_1, vaesdec_<mode>, vaesdeclast_<mode>,
751 vaesenc_<mode>, vpclmulqdq_<mode>,
752 avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): Formatting fixes.
754 2017-12-28 Michael Meissner <meissner@linux.vnet.ibm.com>
756 * builtins.def: (_Float<N> and _Float<N>X BUILT_IN_CEIL): Add
757 _Float<N> and _Float<N>X variants for rounding built-in
759 (_Float<N> and _Float<N>X BUILT_IN_FLOOR): Likewise.
760 (_Float<N> and _Float<N>X BUILT_IN_NEARBYINT): Likewise.
761 (_Float<N> and _Float<N>X BUILT_IN_RINT): Likewise.
762 (_Float<N> and _Float<N>X BUILT_IN_ROUND): Likewise.
763 (_Float<N> and _Float<N>X BUILT_IN_TRUNC): Likewise.
764 * builtins.c (mathfn_built_in_2): Likewise.
765 * internal-fn.def (CEIL): Likewise.
767 (NEARBYINT): Likewise.
771 * convert.c (convert_to_integer_1): Likewise.
772 * fold-const.c (tree_call_nonnegative_warnv_p): Likewise.
773 (integer_valued_real_call_p): Likewise.
774 * fold-const-call.c (fold_const_call_ss): Likewise.
775 * gencfn-macros.c (print_case_cfn): Change CFN and operator
776 printers to take a const char * suffix instead of a bool.
777 (print_define_operator_list): Likewise.
778 (fltall_suffixes): New list of suffixes, that include the
779 traditional suffixes as well as all of the _Float<N> and
781 (main): For _Float<N> and _Float<N>X functions, emit both
782 <name>_FN and <name>_ALL variants. The <macro>_FN variant only
783 has the _Float<N> and _Float<N>X case names or operators. The
784 <name>_ALL variant has both the traditional and the
785 _Float<N>/_Float<N>X case names or operators.
786 * match.pd (COPYSIGN optimizations): Provide optimizations for
787 _Float<N> and _Float<N>X types where possible.
788 (MIN/MAX optimizations): Likewise.
789 (sqrt optimizations): Likewise.
790 (rounding optimizations): Likewise.
792 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
794 * config/powerpcspe/powerpcspe.md (ieee_128bit_negative_zero): Use
795 gen_int_mode rather than GEN_INT.
796 * config/rs6000/rs6000.md (ieee_128bit_negative_zero): Likewise.
798 2017-12-28 Richard Sandiford <richard.sandiford@linaro.org>
800 * emit-rtl.c (gen_const_vec_series): Use valid_for_const_vector_p
801 instead of CONSTANT_P.
802 (gen_vec_series): Likewise.
803 * simplify-rtx.c (simplify_binary_operation_1): Likewise.
805 2017-12-28 Andreas Schwab <schwab@linux-m68k.org>
807 * config/m68k/m68k.md (ashrdi3_const1, lshrdi3_const1): Add
810 2017-12-27 Kugan Vivekanandarajah <kuganv@linaro.org>
812 * config/aarch64/aarch64-simd.md (aarch64_ld1x2<VQ:mode>): New.
813 (aarch64_ld1x2<VDC:mode>): Likewise.
814 (aarch64_simd_ld1<mode>_x2): Likewise.
815 (aarch64_simd_ld1<mode>_x2): Likewise.
816 * config/aarch64/arm_neon.h (vld1_u8_x2): New.
817 (vld1_s8_x2): Likewise.
818 (vld1_u16_x2): Likewise.
819 (vld1_s16_x2): Likewise.
820 (vld1_u32_x2): Likewise.
821 (vld1_s32_x2): Likewise.
822 (vld1_u64_x2): Likewise.
823 (vld1_s64_x2): Likewise.
824 (vld1_f16_x2): Likewise.
825 (vld1_f32_x2): Likewise.
826 (vld1_f64_x2): Likewise.
827 (vld1_p8_x2): Likewise.
828 (vld1_p16_x2): Likewise.
829 (vld1_p64_x2): Likewise.
830 (vld1q_u8_x2): Likewise.
831 (vld1q_s8_x2): Likewise.
832 (vld1q_u16_x2): Likewise.
833 (vld1q_s16_x2): Likewise.
834 (vld1q_u32_x2): Likewise.
835 (vld1q_s32_x2): Likewise.
836 (vld1q_u64_x2): Likewise.
837 (vld1q_s64_x2): Likewise.
838 (vld1q_f16_x2): Likewise.
839 (vld1q_f32_x2): Likewise.
840 (vld1q_f64_x2): Likewise.
841 (vld1q_p8_x2): Likewise.
842 (vld1q_p16_x2): Likewise.
843 (vld1q_p64_x2): Likewise.
845 2017-12-27 Martin Liska <mliska@suse.cz>
847 PR tree-optimization/83552
848 * tree-ssa-strlen.c (fold_strstr_to_strncmp): Assign result
849 of get_string_lenth to a SSA_NAME if not a GIMPLE value.
851 2017-12-27 Tom de Vries <tom@codesourcery.com>
853 * config/nvptx/nvptx.c (nvptx_option_override): Disable
854 -gstatement-frontiers.
856 2017-12-26 Alexander Monakov <amonakov@ispras.ru>
858 PR rtl-optimization/83513
859 * sel-sched.c (sel_rank_for_schedule): Order by non-zero usefulness
860 before priority comparison.
862 2017-12-25 Jakub Jelinek <jakub@redhat.com>
865 * config/i386/i386.opt (-mavx512vpopcntdq, -mmavx512bitalg): Move from
866 ix86_isa_flags2 to ix86_isa_flags.
867 * config/i386/i386-c.c (ix86_target_macros_internal): Test
868 OPTION_MASK_ISA_AVX512BITALG and OPTION_MASK_ISA_AVX512VPOPCNTDQ in
869 isa_flags rather than isa_flags2.
870 * config/i386/i386.c (ix86_target_string): Move -mavx512vpopcntdq
871 and -mavx512bitalg from isa2_opts to isa_opts.
872 (ix86_option_override_internal): Test OPTION_MASK_ISA_AVX512VPOPCNTDQ
873 in x_ix86_isa_flags_explicit rather than x_ix86_isa_flags2_explicit
874 and set it in x_ix86_isa_flags rather than x_ix86_isa_flags2.
876 (def_builtin): Treat OPTION_MASK_ISA_AVX512BW or
877 OPTION_MASK_ISA_AVX512F ored with another option similarly to
878 OPTION_MASK_ISA_AVX512VL. Even for OPTION_MASK_ISA_AVX512VL don't
879 clear it if mask is just OPTION_MASK_ISA_AVX512VL itself.
880 (ix86_expand_builtin): Don't handle OPTION_MASK_ISA_GFNI and
881 OPTION_MASK_ISA_VPCLMULQDQ specially, instead handle
882 OPTION_MASK_ISA_AVX512BW and OPTION_MASK_ISA_AVX512F that way.
883 * config/i386/i386-builtin.def: Move AVX512VPOPCNTDQ and AVX512BITALG
884 builtins from bdesc_args2 to bdesc_args section.
885 (__builtin_ia32_compressstoreuqi512_mask,
886 __builtin_ia32_compressstoreuhi512_mask,
887 __builtin_ia32_compressstoreuqi256_mask,
888 __builtin_ia32_expandloadqi512_mask,
889 __builtin_ia32_expandloadqi512_maskz,
890 __builtin_ia32_expandloadhi512_mask,
891 __builtin_ia32_expandloadhi512_maskz,
892 __builtin_ia32_compressqi512_mask, __builtin_ia32_compresshi512_mask,
893 __builtin_ia32_compressqi256_mask, __builtin_ia32_expandqi512_mask,
894 __builtin_ia32_expandqi512_maskz, __builtin_ia32_expandhi512_mask,
895 __builtin_ia32_expandhi512_maskz, __builtin_ia32_expandqi256_mask,
896 __builtin_ia32_expandqi256_maskz, __builtin_ia32_vpshrd_v32hi_mask,
897 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshrdv_v32hi_mask,
898 __builtin_ia32_vpshrdv_v32hi_maskz, __builtin_ia32_vpshldv_v32hi_mask,
899 __builtin_ia32_vpshldv_v32hi_maskz,
900 __builtin_ia32_vpopcountb_v64qi_mask,
901 __builtin_ia32_vpopcountw_v32hi_mask,
902 __builtin_ia32_vpshufbitqmb512_mask,
903 __builtin_ia32_vpshufbitqmb256_mask): Add
904 " | OPTION_MASK_ISA_AVX512BW".
905 (__builtin_ia32_expandloadqi256_mask,
906 __builtin_ia32_expandloadqi256_maskz,
907 __builtin_ia32_vpopcountb_v32qi_mask): Add
908 " | OPTION_MASK_ISA_AVX512VL | OPTION_MASK_ISA_AVX512BW".
909 (__builtin_ia32_expandloadhi256_mask,
910 __builtin_ia32_expandloadhi256_maskz,
911 __builtin_ia32_expandloadqi128_mask,
912 __builtin_ia32_expandloadqi128_maskz,
913 __builtin_ia32_expandloadhi128_mask,
914 __builtin_ia32_expandloadhi128_maskz,
915 __builtin_ia32_vpshrd_v16hi, __builtin_ia32_vpshrd_v16hi_mask,
916 __builtin_ia32_vpshrd_v8hi, __builtin_ia32_vpshrd_v8hi_mask,
917 __builtin_ia32_vpshrd_v8si, __builtin_ia32_vpshrd_v8si_mask,
918 __builtin_ia32_vpshrd_v4si, __builtin_ia32_vpshrd_v4si_mask,
919 __builtin_ia32_vpshrd_v4di, __builtin_ia32_vpshrd_v4di_mask,
920 __builtin_ia32_vpshrd_v2di, __builtin_ia32_vpshrd_v2di_mask,
921 __builtin_ia32_vpshld_v16hi, __builtin_ia32_vpshld_v16hi_mask,
922 __builtin_ia32_vpshld_v8hi, __builtin_ia32_vpshld_v8hi_mask,
923 __builtin_ia32_vpshld_v8si, __builtin_ia32_vpshld_v8si_mask,
924 __builtin_ia32_vpshld_v4si, __builtin_ia32_vpshld_v4si_mask,
925 __builtin_ia32_vpshld_v4di, __builtin_ia32_vpshld_v4di_mask,
926 __builtin_ia32_vpshld_v2di, __builtin_ia32_vpshld_v2di_mask,
927 __builtin_ia32_vpshrdv_v16hi, __builtin_ia32_vpshrdv_v16hi_mask,
928 __builtin_ia32_vpshrdv_v16hi_maskz, __builtin_ia32_vpshrdv_v8hi,
929 __builtin_ia32_vpshrdv_v8hi_mask, __builtin_ia32_vpshrdv_v8hi_maskz,
930 __builtin_ia32_vpshrdv_v8si, __builtin_ia32_vpshrdv_v8si_mask,
931 __builtin_ia32_vpshrdv_v8si_maskz, __builtin_ia32_vpshrdv_v4si,
932 __builtin_ia32_vpshrdv_v4si_mask, __builtin_ia32_vpshrdv_v4si_maskz,
933 __builtin_ia32_vpshrdv_v4di, __builtin_ia32_vpshrdv_v4di_mask,
934 __builtin_ia32_vpshrdv_v4di_maskz, __builtin_ia32_vpshrdv_v2di,
935 __builtin_ia32_vpshrdv_v2di_mask, __builtin_ia32_vpshrdv_v2di_maskz,
936 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
937 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
938 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
939 __builtin_ia32_vpshldv_v8si, __builtin_ia32_vpshldv_v8si_mask,
940 __builtin_ia32_vpshldv_v8si_maskz, __builtin_ia32_vpshldv_v4si,
941 __builtin_ia32_vpshldv_v4si_mask, __builtin_ia32_vpshldv_v4si_maskz,
942 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
943 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
944 __builtin_ia32_vpshldv_v2di_mask, __builtin_ia32_vpshldv_v2di_maskz,
945 __builtin_ia32_vpopcountb_v32qi, __builtin_ia32_vpopcountb_v16qi,
946 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v16hi,
947 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
948 __builtin_ia32_vpopcountw_v8hi_mask): Add
949 " | OPTION_MASK_ISA_AVX512VL".
950 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
951 _mm512_shrdi_epi32, _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32,
952 _mm512_shrdi_epi64, _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64,
953 _mm512_shldi_epi16, _mm512_shldi_epi32, _mm512_mask_shldi_epi32,
954 _mm512_maskz_shldi_epi32, _mm512_shldi_epi64, _mm512_mask_shldi_epi64,
955 _mm512_maskz_shldi_epi64, _mm512_shrdv_epi16, _mm512_shrdv_epi32,
956 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
957 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64, _mm512_shldv_epi16,
958 _mm512_shldv_epi32, _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32,
959 _mm512_shldv_epi64, _mm512_mask_shldv_epi64,
960 _mm512_maskz_shldv_epi64): Don't require avx512bw for these intrinsics.
961 * config/i386/avx512bitalgintrin.h (_mm_bitshuffle_epi64_mask,
962 _mm_mask_bitshuffle_epi64_mask): Likewise.
963 * common/config/i386/i386-common.c
964 (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
965 OPTION_MASK_ISA_AVX512BITALG_SET): Or in OPTION_MASK_ISA_AVX512F_SET.
966 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
967 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET and
968 OPTION_MASK_ISA_AVX512BITALG_UNSET.
969 (OPTION_MASK_ISA2_AVX512F_UNSET,
970 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Define.
971 (ix86_handle_option): For -mno-general-regs-only, clear from
972 ix86_isa_flags2 OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET rather than
973 just OPTION_MASK_ISA_MPX. For -mno-sse{,2,3,4,4.1,4.2,avx,avx2} and
974 -mno-ssse3 clear OPTION_MASK_ISA2_AVX512F_UNSET bits from
975 ix86_isa_flags2. For -mno-avx512f likewise, instead of masking
976 individually listed ISAs. For -m{,no-}avx512{vpopcntdq,bitalg} adjust
977 for moving from ix86_isa_flags2 to ix86_isa_flags.
979 2017-12-23 Jakub Jelinek <jakub@redhat.com>
982 * fold-const.c (struct contains_label_data): New type.
983 (contains_label_1): Return non-NULL even for CASE_LABEL_EXPR, unless
984 inside of a SWITCH_BODY seen during the walk.
985 (contains_label_p): Use walk_tree instead of
986 walk_tree_without_duplicates, prepare data for contains_label_1 and
989 2017-12-22 Jakub Jelinek <jakub@redhat.com>
992 * tree-iterator.c (alloc_stmt_list): Start with cleared
993 TREE_SIDE_EFFECTS regardless whether a new STATEMENT_LIST is allocated
997 * config/i386/avx512vnniintrin.h: Don't check for __AVX512F__ nor
998 enable avx512f explicitly in #pragma GCC target.
999 * config/i386/i386-builtin.def (__builtin_ia32_vpdpbusd_v8si,
1000 __builtin_ia32_vpdpbusd_v8si_mask, __builtin_ia32_vpdpbusd_v8si_maskz,
1001 __builtin_ia32_vpdpbusd_v4si, __builtin_ia32_vpdpbusd_v4si_mask,
1002 __builtin_ia32_vpdpbusd_v4si_maskz, __builtin_ia32_vpdpbusds_v8si,
1003 __builtin_ia32_vpdpbusds_v8si_mask,
1004 __builtin_ia32_vpdpbusds_v8si_maskz, __builtin_ia32_vpdpbusds_v4si,
1005 __builtin_ia32_vpdpbusds_v4si_mask,
1006 __builtin_ia32_vpdpbusds_v4si_maskz, __builtin_ia32_vpdpwssd_v8si,
1007 __builtin_ia32_vpdpwssd_v8si_mask, __builtin_ia32_vpdpwssd_v8si_maskz,
1008 __builtin_ia32_vpdpwssd_v4si, __builtin_ia32_vpdpwssd_v4si_mask,
1009 __builtin_ia32_vpdpwssd_v4si_maskz, __builtin_ia32_vpdpwssds_v8si,
1010 __builtin_ia32_vpdpwssds_v8si_mask,
1011 __builtin_ia32_vpdpwssds_v8si_maskz, __builtin_ia32_vpdpwssds_v4si,
1012 __builtin_ia32_vpdpwssds_v4si_mask,
1013 __builtin_ia32_vpdpwssds_v4si_maskz): Use
1014 OPTION_MASK_ISA_AVX512VNNI | OPTION_MASK_ISA_AVX512VL instead of
1015 just OPTION_MASK_ISA_AVX512VNNI.
1017 2017-12-22 Martin Jambor <mjambor@suse.cz>
1020 * lto-cgraph.c (output_cgraph_opt_summary_p): Also check former
1023 2017-12-22 Julia Koval <julia.koval@intel.com>
1024 Sebastian Peryt <sebastian.peryt@intel.com>
1026 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512BITALG_SET,
1027 OPTION_MASK_ISA_AVX512BITALG_UNSET): New.
1028 (ix86_handle_option): Handle -mavx512bitalg, fix 4VNNIW formatting.
1029 * config.gcc: Add avx512vpopcntdqvlintrin.h and avx512bitalgintrin.h.
1030 * config/i386/avx512bitalgintrin.h (_mm512_popcnt_epi8, _mm512_popcnt_epi16,
1031 _mm512_mask_popcnt_epi8, _mm512_maskz_popcnt_epi8, _mm512_mask_popcnt_epi16,
1032 _mm512_maskz_popcnt_epi16, _mm512_bitshuffle_epi64_mask, _mm256_popcnt_epi8,
1033 _mm512_mask_bitshuffle_epi64_mask, _mm256_mask_popcnt_epi8, _mm_popcnt_epi8,
1034 _mm256_maskz_popcnt_epi8, _mm_bitshuffle_epi64_mask, _mm256_popcnt_epi16,
1035 _mm_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
1036 _mm256_mask_bitshuffle_epi64_mask, _mm_popcnt_epi16, _mm_maskz_popcnt_epi8,
1037 _mm256_mask_popcnt_epi16, _mm256_maskz_popcnt_epi16, _mm_mask_popcnt_epi8,
1038 _mm_mask_popcnt_epi16, _mm_maskz_popcnt_epi16): New intrinsics.
1039 * config/i386/avx512vpopcntdqvlintrin.h (_mm_popcnt_epi32, _mm_popcnt_epi64,
1040 _mm_mask_popcnt_epi32, _mm_maskz_popcnt_epi32, _mm256_popcnt_epi32,
1041 _mm256_mask_popcnt_epi32, _mm256_maskz_popcnt_epi32, _mm_mask_popcnt_epi64,
1042 _mm_maskz_popcnt_epi64, _mm256_popcnt_epi64, _mm256_mask_popcnt_epi64,
1043 _mm256_maskz_popcnt_epi64): New intrinsics.
1044 * config/i386/cpuid.h (bit_AVX512BITALG): New bit.
1045 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mavx512bitalg.
1046 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI, V64QI_FTYPE_V64QI,
1047 V4DI_FTYPE_V4DI, UHI_FTYPE_V2DI_V2DI_UHI, USI_FTYPE_V4DI_V4DI_USI,
1048 V4SI_FTYPE_V4SI_V4SI_UHI, V8SI_FTYPE_V8SI_V8SI_UHI): New types.
1049 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountq_v4di,
1050 __builtin_ia32_vpopcountq_v4di_mask, __builtin_ia32_vpopcountq_v2di,
1051 __builtin_ia32_vpopcountq_v2di_mask, __builtin_ia32_vpopcountd_v4si,
1052 __builtin_ia32_vpopcountd_v4si_mask, __builtin_ia32_vpopcountd_v8si,
1053 __builtin_ia32_vpopcountd_v8si_mask, __builtin_ia32_vpopcountb_v64qi,
1054 __builtin_ia32_vpopcountb_v64qi_mask, __builtin_ia32_vpopcountb_v32qi,
1055 __builtin_ia32_vpopcountb_v32qi_mask, __builtin_ia32_vpopcountb_v16qi,
1056 __builtin_ia32_vpopcountb_v16qi_mask, __builtin_ia32_vpopcountw_v32hi,
1057 __builtin_ia32_vpopcountw_v32hi_mask, __builtin_ia32_vpopcountw_v16hi,
1058 __builtin_ia32_vpopcountw_v16hi_mask, __builtin_ia32_vpopcountw_v8hi,
1059 __builtin_ia32_vpopcountw_v8hi_mask, __builtin_ia32_vpshufbitqmb128_mask,
1060 __builtin_ia32_vpshufbitqmb256_mask,
1061 __builtin_ia32_vpshufbitqmb512_mask): New builtins.
1062 * config/i386/i386-c.c (__AVX512BITALG__): New.
1063 * config/i386/i386.c (isa2_opts): Add -mavx512bitalg.
1064 (ix86_valid_target_attribute_inner_p): Ditto.
1065 (ix86_expand_args_builtin): Handle new types.
1066 * config/i386/i386.h (TARGET_AVX512BITALG, TARGET_AVX512BITALG_P): New.
1067 * config/i386/i386.opt: Add -mavx512bitalg.
1068 * config/i386/immintrin.h: Add avx512vpopcntdqvlintrin.h and
1069 avx512bitalgintrin.h.
1070 * config/i386/sse.md (VI48_AVX512VLBW): New iterator.
1071 (vpopcount<mode><mask_name>): Add more types.
1072 (avx512vl_vpshufbitqmb<mode><mask_scalar_merge_name>): New.
1073 * doc/invoke.texi: Add -mavx512bitalg and -mavx512vpopcntdq.
1075 2017-12-22 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1077 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET):
1078 Or in OPTION_MASK_ISA_AVX512F_SET.
1079 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1080 OPTION_MASK_ISA_AVX512VNNI_UNSET.
1081 (ix86_handle_option): Adjust for
1082 OPTION_MASK_ISA_AVX512VNNI_*SET being in ix86_isa_flags.
1083 * config/i386/i386-builtin.def: Move VNNI builtins from ARGS2
1085 * config/i386/i386-c.c: Check for OPTION_MASK_ISA_AVX512VNNI in
1086 isa_flag instead of isa_flag2.
1087 * config/i386/i386.c (ix86_target_string): Move -mavx512vnni from
1088 isa_opts2 to isa_opts.
1089 * config/i386/i386.opt (mavx512vnni): Move from ix86_isa_flags2
1092 2017-12-22 Mike Stump <mikestump@comcast.net>
1093 Eric Botcazou <ebotcazou@adacore.com>
1095 * doc/extend.texi (Loop-Specific Pragmas): Document pragma GCC unroll.
1097 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1100 * config/i386/i386.c (ix86_function_arg_boundary): Return
1101 PARM_BOUNDARY for TYPE_EMPTY_P types.
1104 * gimple-ssa-sprintf.c (maybe_warn): Don't call set_caret_index
1105 if navail is >= dir.len.
1107 2017-12-21 Steve Ellcey <sellcey@cavium.com>
1109 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Fix
1112 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1114 PR rtl-optimization/80747
1115 PR rtl-optimization/83512
1116 * cfgrtl.c (force_nonfallthru_and_redirect): When splitting
1117 succ edge from ENTRY, copy partition from e->dest to the newly
1119 * bb-reorder.c (reorder_basic_blocks_simple): If last_tail is
1120 ENTRY, use BB_PARTITION of its successor block as current_partition.
1121 Don't copy partition when splitting succ edge from ENTRY.
1123 PR tree-optimization/83523
1124 * tree-ssa-math-opts.c (is_widening_mult_p): Return false if
1125 for INTEGER_TYPE TYPE_OVERFLOW_TRAPS.
1126 (convert_mult_to_fma): Likewise.
1128 PR tree-optimization/83521
1129 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Use
1130 gimple_build_assign without code on result of
1131 fold_build1 (VIEW_CONVERT_EXPR, ...), as it might not create
1132 a VIEW_CONVERT_EXPR.
1134 2017-12-21 Andrew Pinski <apinski@cavium.com>
1135 Steve Ellcey <sellcey@cavium.com>
1137 * config/aarch64/t-aarch64-linux (MULTILIB_OSDIRNAMES): Handle
1138 multi-arch for ilp32.
1140 2017-12-21 Uros Bizjak <ubizjak@gmail.com>
1143 * config/i386/i386.md (*ashl<mode>3_mask): Add operand
1144 constraints to operand 2.
1145 (*ashl<mode>3_mask_1): Ditto.
1146 (*<shift_insn><mode>3_mask): Ditto.
1147 (*<shift_insn><mode>3_mask_1): Ditto.
1148 (*<rotate_insn><mode>3_mask): Ditto.
1149 (*<rotate_insn><mode>3_mask_1): Ditto.
1151 2017-12-21 Alexandre Oliva <aoliva@redhat.com>
1153 * reorg.c (make_return_insns): Reemit each insn with its own location.
1155 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1157 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
1158 generation for cases where splatting a value is not useful.
1159 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1160 across a vec_duplicate and a paradoxical subreg forming a vector
1161 mode to a vec_concat.
1163 2017-12-21 James Greenhalgh <james.greenhalgh@arm.com>
1165 * combine.c (simplify_set): Do not transform subregs to zero_extends
1166 if the destination is not a scalar int mode.
1168 2017-12-21 Eric Botcazou <ebotcazou@adacore.com>
1171 * convert.c (convert_to_integer_1) <POINTER_TYPE>: Do not return the
1172 shared zero if the input has overflowed.
1174 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1176 * config/arm/driver-arm.c (arm_cpu_table): Specify dotprod
1177 support for Cortex-A55 and Cortex-A75.
1179 2017-12-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1181 * common/config/arm/arm-common.c (compare_opt_names): Add function
1182 comment. Use strcmp instead of manual loop.
1184 2017-12-21 Martin Liska <mliska@suse.cz>
1186 PR gcov-profile/83509
1187 * gcov-dump.c (dump_gcov_file): Do not read info about
1188 support_unexecuted_blocks for gcda files.
1190 2017-12-21 Jakub Jelinek <jakub@redhat.com>
1192 PR rtl-optimization/82973
1193 * emit-rtl.h (valid_for_const_vec_duplicate_p): Rename to ...
1194 (valid_for_const_vector_p): ... this.
1195 * emit-rtl.c (valid_for_const_vec_duplicate_p): Rename to ...
1196 (valid_for_const_vector_p): ... this. Adjust function comment.
1197 (gen_vec_duplicate): Adjust caller.
1198 * optabs.c (expand_vector_broadcast): Likewise.
1199 * simplify-rtx.c (simplify_const_unary_operation): Don't optimize into
1200 CONST_VECTOR if some element isn't simplified valid_for_const_vector_p
1202 (simplify_const_binary_operation): Likewise. Use CONST_FIXED_P macro
1203 instead of GET_CODE == CONST_FIXED.
1204 (simplify_subreg): Use CONST_FIXED_P macro instead of
1205 GET_CODE == CONST_FIXED.
1208 * config/i386/i386.c (ix86_target_string): Move -mavx512vbmi2 and
1209 -mshstk entries from isa_opts2 to isa_opts and -mhle, -mmovbe,
1210 -mclzero and -mmwaitx entries from isa_opts to isa_opts2.
1211 (ix86_option_override_internal): Adjust for
1212 OPTION_MASK_ISA_{HLE,MOVBE,CLZERO,MWAITX} moving to ix86_isa_flags2
1213 and OPTION_MASK_ISA_SHSTK moving to ix86_isa_flags.
1214 (BDESC_VERIFYS): Remove SPECIAL_ARGS2 related checks.
1215 (ix86_init_mmx_sse_builtins): Remove bdesc_special_args2 handling.
1216 Use def_builtin2 instead of def_builtin for OPTION_MASK_ISA_MWAITX
1217 and OPTION_MASK_ISA_CLZERO builtins. Use def_builtin instead of
1218 def_builtin2 for CET builtins.
1219 (ix86_expand_builtin): Remove bdesc_special_args2 handling. Fix
1220 up formatting in IX86_BUILTIN_RDPID code.
1221 * config/i386/i386-builtin.def: Move VBMI2 builtins from SPECIAL_ARGS2
1222 section to SPECIAL_ARGS and from ARGS2 section to ARGS.
1223 * config/i386/i386.opt (mavx512vbmi2, mshstk): Move from
1224 ix86_isa_flags2 to ix86_isa_flags.
1225 (mhle, mmovbe, mclzero, mmwaitx): Move from ix86_isa_flags to
1227 * config/i386/i386-c.c (ix86_target_macros_internal): Check for
1228 OPTION_MASK_ISA_{CLZERO,MWAITX} in isa_flag2 instead of isa_flag.
1229 Check for OPTION_MASK_ISA_{SHSTK,AVX512VBMI2} in isa_flag instead
1231 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET):
1232 Or in OPTION_MASK_ISA_AVX512F_SET.
1233 (OPTION_MASK_ISA_AVX512F_UNSET): Or in
1234 OPTION_MASK_ISA_AVX512VBMI2_UNSET.
1235 (ix86_handle_option): Adjust for
1236 OPTION_MASK_ISA_{SHSTK,AVX512VBMI2}_*SET being in ix86_isa_flags
1237 and OPTION_MASK_ISA_{MOVBE,MWAITX,CLZERO}_*SET in ix86_isa_flags2.
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-data-ref.h (prune_runtime_alias_test_list): Take the
1244 factor as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1245 * tree-data-ref.c (prune_runtime_alias_test_list): Likewise.
1246 Track polynomial offsets.
1248 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1249 Alan Hayward <alan.hayward@arm.com>
1250 David Sherwood <david.sherwood@arm.com>
1252 * tree-vect-data-refs.c (vect_compute_data_ref_alignment):
1253 Treat drb->init as a poly_int. Fail if its misalignment wrt
1254 vector_alignment isn't known.
1256 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1257 Alan Hayward <alan.hayward@arm.com>
1258 David Sherwood <david.sherwood@arm.com>
1260 * tree-vectorizer.h (_loop_vec_info): Add a versioning_threshold
1262 (LOOP_VINFO_VERSIONING_THRESHOLD): New macro
1263 (vect_loop_versioning): Take the loop versioning threshold as a
1265 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
1266 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
1267 versioning_threshold.
1268 (vect_analyze_loop_2): Compute the loop versioning threshold
1269 whenever loop versioning is needed, and store it in the new
1270 field rather than combining it with the cost model threshold.
1271 (vect_transform_loop): Update call to vect_loop_versioning.
1272 Try to combine the loop versioning and cost thresholds here.
1274 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1275 Alan Hayward <alan.hayward@arm.com>
1276 David Sherwood <david.sherwood@arm.com>
1278 * tree-ssa-loop-ivopts.h (strip_offset): Return the offset as
1279 poly_uint64_pod rather than an unsigned HOST_WIDE_INT.
1280 * tree-loop-distribution.c (classify_builtin_st): Update accordingly.
1281 * tree-ssa-loop-ivopts.c (iv_use::addr_offset): Change from
1282 an unsigned HOST_WIDE_INT to a poly_uint64_pod.
1283 (group_compare_offset): Update accordingly.
1284 (split_small_address_groups_p): Likewise.
1285 (record_use): Take addr_offset as a poly_uint64 rather than
1286 an unsigned HOST_WIDE_INT.
1287 (strip_offset): Return the offset as a poly_uint64 rather than
1288 an unsigned HOST_WIDE_INT.
1289 (record_group_use, split_address_groups): Track polynomial offsets.
1290 (add_iv_candidate_for_use): Likewise.
1291 (addr_offset_valid_p): Take the offset as a poly_int64 rather
1292 than a HOST_WIDE_INT.
1293 (strip_offset_1): Return the offset as a poly_int64 rather than
1296 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1297 Alan Hayward <alan.hayward@arm.com>
1298 David Sherwood <david.sherwood@arm.com>
1300 * tree.h (get_binfo_at_offset): Take the offset as a poly_int64
1301 rather than a HOST_WIDE_INT.
1302 * tree.c (get_binfo_at_offset): Likewise.
1304 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1305 Alan Hayward <alan.hayward@arm.com>
1306 David Sherwood <david.sherwood@arm.com>
1308 * ipa-prop.h (build_ref_for_offset): Take the offset as a poly_int64
1309 rather than a HOST_WIDE_INT.
1310 * tree-sra.c (build_ref_for_offset): Likewise.
1312 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1313 Alan Hayward <alan.hayward@arm.com>
1314 David Sherwood <david.sherwood@arm.com>
1316 * fold-const.h (mem_ref_offset): Return a poly_offset_int rather
1318 * tree.c (mem_ref_offset): Likewise.
1319 (build_simple_mem_ref_loc): Treat MEM_REF offsets as poly_ints.
1320 * builtins.c (get_object_alignment_2): Likewise.
1321 * expr.c (get_inner_reference, expand_expr_real_1): Likewise.
1322 * gimple-fold.c (get_base_constructor): Likewise.
1323 * gimple-ssa-strength-reduction.c (restructure_reference): Likewise.
1324 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1326 * ipa-polymorphic-call.c
1327 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
1328 * ipa-prop.c (compute_complex_assign_jump_func): Likewise.
1329 (get_ancestor_addr_info): Likewise.
1330 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Likewise.
1331 * match.pd: Likewise.
1332 * tree-data-ref.c (dr_analyze_innermost): Likewise.
1333 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1334 * tree-eh.c (tree_could_trap_p): Likewise.
1335 * tree-object-size.c (addr_object_size): Likewise.
1336 * tree-ssa-address.c (copy_ref_info): Likewise.
1337 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p): Likewise.
1338 (indirect_refs_may_alias_p): Likewise.
1339 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1340 * tree-ssa.c (maybe_rewrite_mem_ref_base): Likewise.
1341 (non_rewritable_mem_ref_base): Likewise.
1342 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1343 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1344 * varasm.c (decode_addr_const): Likewise.
1346 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1347 Alan Hayward <alan.hayward@arm.com>
1348 David Sherwood <david.sherwood@arm.com>
1350 * gimple-ssa-stor-merging.c (find_bswap_or_nop_load): Track polynomial
1351 offsets for MEM_REFs.
1353 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1354 Alan Hayward <alan.hayward@arm.com>
1355 David Sherwood <david.sherwood@arm.com>
1357 * tree-ssanames.h (adjust_ptr_info_misalignment): Take the increment
1358 as a poly_uint64 rather than an unsigned int.
1359 * tree-ssanames.c (adjust_ptr_info_misalignment): Likewise.
1361 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1362 Alan Hayward <alan.hayward@arm.com>
1363 David Sherwood <david.sherwood@arm.com>
1365 * varasm.c (addr_const::offset): Change from HOST_WIDE_INT
1367 (decode_addr_const): Update accordingly.
1369 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1370 Alan Hayward <alan.hayward@arm.com>
1371 David Sherwood <david.sherwood@arm.com>
1373 * tree.h (bit_field_size, bit_field_offset): New functions.
1374 * hsa-gen.c (gen_hsa_addr): Use them.
1375 * tree-ssa-forwprop.c (simplify_bitfield_ref): Likewise.
1376 (simplify_vector_constructor): Likewise.
1377 * tree-ssa-sccvn.c (copy_reference_ops_from_ref): Likewise.
1378 * tree-cfg.c (verify_expr): Require the sizes and offsets of a
1379 BIT_FIELD_REF to be poly_uint64s rather than uhwis.
1380 * fold-const.c (fold_ternary_loc): Protect tree_to_uhwi with
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 * expr.h (emit_group_load, emit_group_load_into_temps)
1388 (emit_group_store): Take the size as a poly_int64 rather than an int.
1389 * expr.c (emit_group_load_1, emit_group_load): Likewise.
1390 (emit_group_load_into_temp, emit_group_store): Likewise.
1392 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1393 Alan Hayward <alan.hayward@arm.com>
1394 David Sherwood <david.sherwood@arm.com>
1396 * ira-int.h (ira_spilled_reg_stack_slot::width): Change from
1397 an unsigned int to a poly_uint64.
1398 * ira.h (ira_reuse_stack_slot, ira_mark_new_stack_slot): Take the
1399 sizes as poly_uint64s rather than unsigned ints.
1400 * ira-color.c (ira_reuse_stack_slot, ira_mark_new_stack_slot):
1403 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1404 Alan Hayward <alan.hayward@arm.com>
1405 David Sherwood <david.sherwood@arm.com>
1407 * lra-constraints.c (emit_inc): Change inc_amount from an int
1410 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1411 Alan Hayward <alan.hayward@arm.com>
1412 David Sherwood <david.sherwood@arm.com>
1414 * cfgexpand.c (stack_var::size): Change from a HOST_WIDE_INT
1416 (add_stack_var, stack_var_cmp, partition_stack_vars)
1417 (dump_stack_var_partition): Update accordingly.
1418 (alloc_stack_frame_space): Take the size as a poly_int64 rather
1419 than a HOST_WIDE_INT.
1420 (expand_stack_vars, expand_one_stack_var_1): Handle polynomial sizes.
1421 (defer_stack_allocation, estimated_stack_frame_size): Likewise.
1422 (account_stack_vars, expand_one_var): Likewise. Return a poly_uint64
1423 rather than a HOST_WIDE_INT.
1425 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1426 Alan Hayward <alan.hayward@arm.com>
1427 David Sherwood <david.sherwood@arm.com>
1429 * target.def (return_pops_args): Treat both the input and output
1430 sizes as poly_int64s rather than HOST_WIDE_INTS.
1431 * targhooks.h (default_return_pops_args): Update accordingly.
1432 * targhooks.c (default_return_pops_args): Likewise.
1433 * doc/tm.texi: Regenerate.
1434 * emit-rtl.h (incoming_args): Change pops_args, size and
1435 outgoing_args_size from int to poly_int64_pod.
1436 * function.h (expr_status): Change x_pending_stack_adjust and
1437 x_stack_pointer_delta from int to poly_int64.
1438 (args_size::constant): Change from HOST_WIDE_INT to poly_int64.
1439 (ARGS_SIZE_RTX): Update accordingly.
1440 * calls.c (highest_outgoing_arg_in_use): Change from int to
1442 (stack_usage_watermark, stored_args_watermark): New variables.
1443 (stack_region_maybe_used_p, mark_stack_region_used): New functions.
1444 (emit_call_1): Change the stack_size and rounded_stack_size
1445 parameters from HOST_WIDE_INT to poly_int64. Track n_popped
1447 (save_fixed_argument_area): Check stack_usage_watermark.
1448 (initialize_argument_information): Change old_pending_adj from
1449 a HOST_WIDE_INT * to a poly_int64_pod *.
1450 (compute_argument_block_size): Return the size as a poly_int64
1452 (finalize_must_preallocate): Track polynomial argument sizes.
1453 (compute_argument_addresses): Likewise.
1454 (internal_arg_pointer_based_exp): Track polynomial offsets.
1455 (mem_overlaps_already_clobbered_arg_p): Rename to...
1456 (mem_might_overlap_already_clobbered_arg_p): ...this and take the
1457 size as a poly_uint64 rather than an unsigned HOST_WIDE_INT.
1458 Check stored_args_used_watermark.
1459 (load_register_parameters): Update accordingly.
1460 (check_sibcall_argument_overlap_1): Likewise.
1461 (combine_pending_stack_adjustment_and_call): Take the unadjusted
1462 args size as a poly_int64 rather than an int. Return a bool
1463 indicating whether the optimization was possible and return
1464 the new adjustment by reference.
1465 (check_sibcall_argument_overlap): Track polynomail argument sizes.
1466 Update stored_args_watermark.
1467 (can_implement_as_sibling_call_p): Handle polynomial argument sizes.
1468 (expand_call): Likewise. Maintain stack_usage_watermark and
1469 stored_args_watermark. Update calls to
1470 combine_pending_stack_adjustment_and_call.
1471 (emit_library_call_value_1): Handle polynomial argument sizes.
1472 Call stack_region_maybe_used_p and mark_stack_region_used.
1473 Maintain stack_usage_watermark.
1474 (store_one_arg): Likewise. Update call to
1475 mem_overlaps_already_clobbered_arg_p.
1476 * config/arm/arm.c (arm_output_function_prologue): Add a cast to
1478 * config/avr/avr.c (avr_outgoing_args_size): Likewise.
1479 * config/microblaze/microblaze.c (microblaze_function_prologue):
1481 * config/cr16/cr16.c (cr16_return_pops_args): Update for new
1482 TARGET_RETURN_POPS_ARGS interface.
1483 (cr16_compute_frame, cr16_initial_elimination_offset): Add casts
1485 * config/ft32/ft32.c (ft32_compute_frame): Likewise.
1486 * config/i386/i386.c (ix86_return_pops_args): Update for new
1487 TARGET_RETURN_POPS_ARGS interface.
1488 (ix86_expand_split_stack_prologue): Add a cast to HOST_WIDE_INT.
1489 * config/moxie/moxie.c (moxie_compute_frame): Likewise.
1490 * config/m68k/m68k.c (m68k_return_pops_args): Update for new
1491 TARGET_RETURN_POPS_ARGS interface.
1492 * config/vax/vax.c (vax_return_pops_args): Likewise.
1493 * config/pa/pa.h (STACK_POINTER_OFFSET): Add a cast to poly_int64.
1494 (EXIT_IGNORE_STACK): Update reference to crtl->outgoing_args_size.
1495 * config/arm/arm.h (CALLER_INTERWORKING_SLOT_SIZE): Likewise.
1496 * config/powerpcspe/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1497 * config/powerpcspe/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1498 * config/powerpcspe/powerpcspe.h (STACK_DYNAMIC_OFFSET): Likewise.
1499 * config/rs6000/aix.h (STACK_DYNAMIC_OFFSET): Likewise.
1500 * config/rs6000/darwin.h (STACK_DYNAMIC_OFFSET): Likewise.
1501 * config/rs6000/rs6000.h (STACK_DYNAMIC_OFFSET): Likewise.
1502 * dojump.h (saved_pending_stack_adjust): Change x_pending_stack_adjust
1503 and x_stack_pointer_delta from int to poly_int64.
1504 * dojump.c (do_pending_stack_adjust): Update accordingly.
1505 * explow.c (allocate_dynamic_stack_space): Handle polynomial
1506 stack_pointer_deltas.
1507 * function.c (STACK_DYNAMIC_OFFSET): Add a cast to poly_int64.
1508 (pad_to_arg_alignment): Track polynomial offsets.
1509 (assign_parm_find_stack_rtl): Likewise.
1510 (assign_parms, locate_and_pad_parm): Handle polynomial argument sizes.
1511 * toplev.c (output_stack_usage): Update reference to
1512 current_function_pushed_stack_size.
1514 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1515 Alan Hayward <alan.hayward@arm.com>
1516 David Sherwood <david.sherwood@arm.com>
1518 * function.c (in_arg_offset, var_offset, dynamic_offset)
1519 (out_arg_offset, cfa_offset): Change from int to poly_int64.
1520 (instantiate_new_reg): Return the new offset as a poly_int64_pod
1521 rather than a HOST_WIDE_INT.
1522 (instantiate_virtual_regs_in_rtx): Track polynomial offsets.
1523 (instantiate_virtual_regs_in_insn): Likewise.
1525 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1526 Alan Hayward <alan.hayward@arm.com>
1527 David Sherwood <david.sherwood@arm.com>
1529 * rtl.h (get_args_size, add_args_size_note): New functions.
1530 (find_args_size_adjust): Return a poly_int64 rather than a
1532 (fixup_args_size_notes): Likewise. Make the same change to the
1533 end_args_size parameter.
1534 * rtlanal.c (get_args_size, add_args_size_note): New functions.
1535 * builtins.c (expand_builtin_trap): Use add_args_size_note.
1536 * calls.c (emit_call_1): Likewise.
1537 * explow.c (adjust_stack_1): Likewise.
1538 * cfgcleanup.c (old_insns_match_p): Update use of
1539 find_args_size_adjust.
1540 * combine.c (distribute_notes): Track polynomial arg sizes.
1541 * dwarf2cfi.c (dw_trace_info): Change beg_true_args_size,
1542 end_true_args_size, beg_delay_args_size and end_delay_args_size
1543 from HOST_WIDE_INT to poly_int64.
1544 (add_cfi_args_size): Take the args_size as a poly_int64 rather
1545 than a HOST_WIDE_INT.
1546 (notice_args_size, notice_eh_throw, maybe_record_trace_start)
1547 (maybe_record_trace_start_abnormal, scan_trace, connect_traces): Track
1548 polynomial arg sizes.
1549 * emit-rtl.c (try_split): Use get_args_size.
1550 * recog.c (peep2_attempt): Likewise.
1551 * reload1.c (reload_as_needed): Likewise.
1552 * expr.c (find_args_size_adjust): Return the adjustment as a
1553 poly_int64 rather than a HOST_WIDE_INT.
1554 (fixup_args_size_notes): Change end_args_size from a HOST_WIDE_INT
1555 to a poly_int64 and change the return type in the same way.
1556 (emit_single_push_insn): Track polynomial arg sizes.
1558 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1559 Alan Hayward <alan.hayward@arm.com>
1560 David Sherwood <david.sherwood@arm.com>
1562 * expr.h (push_block, emit_push_insn): Change the "extra" parameter
1563 from HOST_WIDE_INT to poly_int64.
1564 * expr.c (push_block, emit_push_insn): Likewise.
1566 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1567 Alan Hayward <alan.hayward@arm.com>
1568 David Sherwood <david.sherwood@arm.com>
1570 * function.h (frame_space): Change start and length from HOST_WIDE_INT
1572 (get_frame_size): Return the size as a poly_int64 rather than a
1574 (frame_offset_overflow): Take the offset as a poly_int64 rather
1575 than a HOST_WIDE_INT.
1576 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1577 (assign_stack_temp): Likewise for the size.
1578 * function.c (get_frame_size): Return a poly_int64 rather than
1580 (frame_offset_overflow): Take the offset as a poly_int64 rather
1581 than a HOST_WIDE_INT.
1582 (try_fit_stack_local): Take the start, length and size as poly_int64s
1583 rather than HOST_WIDE_INTs. Return the offset as a poly_int64_pod
1584 rather than a HOST_WIDE_INT.
1585 (add_frame_space): Take the start and end as poly_int64s rather than
1587 (assign_stack_local_1, assign_stack_local, assign_stack_temp_for_type)
1588 (assign_stack_temp): Likewise for the size.
1589 (temp_slot): Change size, base_offset and full_size from HOST_WIDE_INT
1591 (find_temp_slot_from_address): Handle polynomial offsets.
1592 (combine_temp_slots): Likewise.
1593 * emit-rtl.h (rtl_data::x_frame_offset): Change from HOST_WIDE_INT
1595 * cfgexpand.c (alloc_stack_frame_space): Return the offset as a
1596 poly_int64 rather than a HOST_WIDE_INT.
1597 (expand_one_stack_var_at): Take the offset as a poly_int64 rather
1598 than a HOST_WIDE_INT.
1599 (expand_stack_vars, expand_one_stack_var_1, expand_used_vars): Handle
1600 polynomial frame offsets.
1601 * config/m32r/m32r-protos.h (m32r_compute_frame_size): Take the size
1602 as a poly_int64 rather than an int.
1603 * config/m32r/m32r.c (m32r_compute_frame_size): Likewise.
1604 * config/v850/v850-protos.h (compute_frame_size): Likewise.
1605 * config/v850/v850.c (compute_frame_size): Likewise.
1606 * config/xtensa/xtensa-protos.h (compute_frame_size): Likewise.
1607 * config/xtensa/xtensa.c (compute_frame_size): Likewise.
1608 * config/pa/pa-protos.h (pa_compute_frame_size): Likewise.
1609 * config/pa/pa.c (pa_compute_frame_size): Likewise.
1610 * explow.h (get_dynamic_stack_base): Take the offset as a poly_int64
1611 rather than a HOST_WIDE_INT.
1612 * explow.c (get_dynamic_stack_base): Likewise.
1613 * final.c (final_start_function): Use the constant lower bound
1614 of the frame size for -Wframe-larger-than.
1615 * ira.c (do_reload): Adjust for new get_frame_size return type.
1616 * lra.c (lra): Likewise.
1617 * reload1.c (reload): Likewise.
1618 * config/avr/avr.c (avr_asm_function_end_prologue): Likewise.
1619 * config/pa/pa.h (EXIT_IGNORE_STACK): Likewise.
1620 * rtlanal.c (get_initial_register_offset): Return the offset as
1621 a poly_int64 rather than a HOST_WIDE_INT.
1623 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1624 Alan Hayward <alan.hayward@arm.com>
1625 David Sherwood <david.sherwood@arm.com>
1627 * reload1.c (elim_table): Change initial_offset, offset and
1628 previous_offset from HOST_WIDE_INT to poly_int64_pod.
1629 (offsets_at): Change the target array's element type from
1630 HOST_WIDE_INT to poly_int64_pod.
1631 (set_label_offsets, eliminate_regs_1, eliminate_regs_in_insn)
1632 (elimination_costs_in_insn, update_eliminable_offsets)
1633 (verify_initial_elim_offsets, set_offsets_for_label)
1634 (init_eliminable_invariants): Update after above changes.
1636 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1637 Alan Hayward <alan.hayward@arm.com>
1638 David Sherwood <david.sherwood@arm.com>
1640 * reload.h (reload::inc): Change from an int to a poly_int64_pod.
1641 * reload.c (combine_reloads, debug_reload_to_stream): Likewise.
1642 (decomposition): Change start and end from HOST_WIDE_INT
1644 (decompose, immune_p): Update accordingly.
1645 (find_inc_amount): Return a poly_int64 rather than an int.
1646 * reload1.c (inc_for_reload): Take the inc_amount as a poly_int64
1649 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1650 Alan Hayward <alan.hayward@arm.com>
1651 David Sherwood <david.sherwood@arm.com>
1653 * tree.h (get_inner_reference): Return the bitsize and bitpos
1654 as poly_int64_pods rather than HOST_WIDE_INT.
1655 * fold-const.h (ptr_difference_const): Return the pointer difference
1656 as a poly_int64_pod rather than a HOST_WIDE_INT.
1657 * expr.c (get_inner_reference): Return the bitsize and bitpos
1658 as poly_int64_pods rather than HOST_WIDE_INT.
1659 (expand_expr_addr_expr_1, expand_expr_real_1): Track polynomial
1661 * fold-const.c (make_bit_field_ref): Take the bitpos as a poly_int64
1662 rather than a HOST_WIDE_INT. Update call to get_inner_reference.
1663 (optimize_bit_field_compare): Update call to get_inner_reference.
1664 (decode_field_reference): Likewise.
1665 (fold_unary_loc): Track polynomial offsets and sizes.
1666 (split_address_to_core_and_offset): Return the bitpos as a
1667 poly_int64_pod rather than a HOST_WIDE_INT.
1668 (ptr_difference_const): Likewise for the pointer difference.
1669 * asan.c (instrument_derefs): Track polynomial offsets and sizes.
1670 * config/mips/mips.c (r10k_safe_mem_expr_p): Likewise.
1671 * dbxout.c (dbxout_expand_expr): Likewise.
1672 * dwarf2out.c (loc_list_for_address_of_addr_expr_of_indirect_ref)
1673 (loc_list_from_tree_1, fortran_common): Likewise.
1674 * gimple-laddress.c (pass_laddress::execute): Likewise.
1675 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Likewise.
1676 * gimplify.c (gimplify_scan_omp_clauses): Likewise.
1677 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
1678 * tree-affine.c (tree_to_aff_combination): Likewise.
1679 (get_inner_reference_aff): Likewise.
1680 * tree-data-ref.c (split_constant_offset_1): Likewise.
1681 (dr_analyze_innermost): Likewise.
1682 * tree-scalar-evolution.c (interpret_rhs_expr): Likewise.
1683 * tree-sra.c (ipa_sra_check_caller): Likewise.
1684 * tree-vect-data-refs.c (vect_check_gather_scatter): Likewise.
1685 * ubsan.c (maybe_instrument_pointer_overflow): Likewise.
1686 (instrument_bool_enum_load, instrument_object_size): Likewise.
1687 * gimple-ssa-strength-reduction.c (slsr_process_ref): Update call
1688 to get_inner_reference.
1689 * hsa-gen.c (gen_hsa_addr): Likewise.
1690 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Likewise.
1691 * tsan.c (instrument_expr): Likewise.
1692 * match.pd: Update call to ptr_difference_const.
1694 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1695 Alan Hayward <alan.hayward@arm.com>
1696 David Sherwood <david.sherwood@arm.com>
1698 * fold-const.c (fold_comparison): Track sizes and offsets as
1699 poly_int64s rather than HOST_WIDE_INTs when folding address
1702 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1703 Alan Hayward <alan.hayward@arm.com>
1704 David Sherwood <david.sherwood@arm.com>
1706 * expr.h (get_bit_range): Return the bitstart and bitend as
1707 poly_uint64s rather than unsigned HOST_WIDE_INTs. Return the bitpos
1708 as a poly_int64 rather than a HOST_WIDE_INT.
1709 * expr.c (get_bit_range): Likewise.
1710 (expand_assignment): Update call accordingly.
1711 * fold-const.c (optimize_bit_field_compare): Likewise.
1713 2017-12-21 Richard Sandiford <richard.sandiford@linaro.org>
1714 Alan Hayward <alan.hayward@arm.com>
1715 David Sherwood <david.sherwood@arm.com>
1717 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): New enum.
1718 (aarch64_legitimate_address_p): Use it instead of an rtx code,
1719 as an optional final parameter.
1720 * config/aarch64/aarch64.c (aarch64_classify_address): Likewise.
1721 (aarch64_legitimate_address_p): Likewise.
1722 (aarch64_print_address_internal): Take an aarch64_addr_query_type
1723 instead of an rtx code.
1724 (aarch64_address_valid_for_prefetch_p): Update calls accordingly.
1725 (aarch64_legitimate_address_hook_p): Likewise.
1726 (aarch64_print_ldpstp_address): Likewise.
1727 (aarch64_print_operand_address): Likewise.
1728 (aarch64_address_cost): Likewise.
1729 * config/aarch64/constraints.md (Uml, Umq, Ump, Utq): Likewise.
1730 * config/aarch64/predicates.md (aarch64_mem_pair_operand): Likewise.
1731 (aarch64_mem_pair_lanes_operand): Likewise.
1733 2017-12-20 Richard Biener <rguenther@suse.de>
1735 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Call
1736 update_stmt_if_modified.
1738 2017-12-20 Wilco Dijkstra <wdijkstr@arm.com>
1740 PR tree-optimization/83491
1741 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Check for SSA_NAME
1742 before walking uses. Improve coding style and comments.
1744 2017-12-20 Tom de Vries <tom@codesourcery.com>
1746 * gimple-fold.c (fold_internal_goacc_dim): Simplify.
1748 2017-12-20 Jakub Jelinek <jakub@redhat.com>
1751 * ipa-fnsummary.c (pass_data_ipa_free_fn_summary): Use 0 for
1753 (pass_ipa_free_fn_summary): Add small_p private data member,
1754 initialize to false in the ctor.
1755 (pass_ipa_free_fn_summary::clone,
1756 pass_ipa_free_fn_summary::set_pass_param,
1757 pass_ipa_free_fn_summary::gate): New methods.
1758 (pass_ipa_free_fn_summary::execute): Return TODO_remove_functions
1759 | TODO_dump_symtab if small_p.
1760 * passes.def: Add true parm for the existing pass_ipa_free_fn_summary
1761 entry and add another instance of the pass with false parm after
1763 * ipa-pure-const.c (pass_ipa_pure_const): Don't call
1764 ipa_free_fn_summary here.
1766 2017-12-20 Paolo Carlini <paolo.carlini@oracle.com>
1768 * gimplify.c (gimplify_return_expr): Remove dead error_mark_node check.
1770 2017-12-20 Martin Sebor <msebor@redhat.com>
1773 * builtins.c (expand_builtin_strlen): Use get_callee_fndecl.
1774 (expand_builtin_strcmp): Call maybe_warn_nonstring_arg.
1775 (expand_builtin_strncmp): Same.
1777 2017-12-20 Alexandre Oliva <aoliva@redhat.com>
1780 * cfgexpand.c (label_rtx_for_bb): Revert SFN changes that
1781 allowed debug stmts before labels.
1782 (expand_gimple_basic_block): Likewise.
1783 * gimple-iterator.c (gimple_find_edge_insert_loc): Likewise.
1784 * gimple-iterator.h (gsi_after_labels): Likewise.
1785 * tree-cfgcleanup (remove_forwarder_block): Likewise, but
1786 rename reused variable, and simplify using gsi_move_before.
1787 * tree-ssa-tail-merge.c (find_duplicate): Likewise.
1788 * tree-cfg.c (make_edges, cleanup_dead_labels): Likewise.
1789 (gimple_can_merge_blocks_p, verify_gimple_in_cfg): Likewise.
1790 (gimple_verify_flow_info, gimple_block_label): Likewise.
1791 (make_blocks): Move debug markers after adjacent labels.
1792 * cfgrtl.c (skip_insns_after_block): Revert SFN changes that
1793 allowed debug insns outside blocks.
1794 * df-scan.c (df_insn_delete): Likewise.
1795 * lra-constraints.c (update_ebb_live_info): Likewise.
1796 * var-tracking.c (get_first_insn, vt_emit_notes): Likewise.
1797 (vt_initialize, delete_vta_debug_insns): Likewise.
1798 (reemit_marker_as_note): Drop BB parm. Adjust callers.
1800 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1801 Alan Hayward <alan.hayward@arm.com>
1802 David Sherwood <david.sherwood@arm.com>
1804 * poly-int-types.h (round_down_to_byte_boundary): New macro.
1805 (round_up_to_byte_boundary): Likewise.
1806 * expr.h (get_bit_range): Add temporary shim.
1807 * gimple-ssa-store-merging.c (store_operand_info): Change the
1808 bitsize, bitpos, bitregion_start and bitregion_end fields from
1809 unsigned HOST_WIDE_INT to poly_uint64.
1810 (merged_store_group): Likewise load_align_base.
1811 (compatible_load_p, compatible_load_p): Update accordingly.
1812 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
1813 (split_group, imm_store_chain_info::output_merged_store): Likewise.
1814 (mem_valid_for_store_merging): Return the bitsize, bitpos,
1815 bitregion_start and bitregion_end as poly_uint64s rather than
1816 unsigned HOST_WIDE_INTs. Track polynomial offsets internally.
1817 (handled_load): Take the bitsize, bitpos,
1818 bitregion_start and bitregion_end as poly_uint64s rather than
1819 unsigned HOST_WIDE_INTs.
1820 (pass_store_merging::process_store): Update call to
1821 mem_valid_for_store_merging.
1823 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1824 Alan Hayward <alan.hayward@arm.com>
1825 David Sherwood <david.sherwood@arm.com>
1827 * builtins.c (get_object_alignment_2): Track polynomial offsets
1828 and sizes. Update the alignment handling.
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.h (get_inner_reference): Add a version that returns the
1835 offset and size as poly_int64_pods rather than HOST_WIDE_INTs.
1836 * cfgexpand.c (expand_debug_expr): Track polynomial offsets. Simply
1837 the case in which bitpos is not associated with the first byte.
1839 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1840 Alan Hayward <alan.hayward@arm.com>
1841 David Sherwood <david.sherwood@arm.com>
1843 * tree-affine.h (get_inner_reference_aff): Return the size as a
1845 * tree-affine.c (get_inner_reference_aff): Likewise.
1846 * tree-data-ref.c (dr_may_alias_p): Update accordingly.
1847 * tree-ssa-loop-im.c (mem_refs_may_alias_p): Likewise.
1849 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1850 Alan Hayward <alan.hayward@arm.com>
1851 David Sherwood <david.sherwood@arm.com>
1853 * fold-const.c (pointer_may_wrap_p): Take the offset as a
1854 HOST_WIDE_INT rather than a poly_int64.
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 * gimple-ssa-store-merging.c (symbolic_number::bytepos): Change from
1861 HOST_WIDE_INT to poly_int64_pod.
1862 (perform_symbolic_merge): Update accordingly.
1863 (bswap_replace): Likewise.
1865 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1866 Alan Hayward <alan.hayward@arm.com>
1867 David Sherwood <david.sherwood@arm.com>
1869 * tree-affine.h (aff_tree::offset): Change from widest_int
1871 (wide_int_ext_for_comb): Delete.
1872 (aff_combination_const, aff_comb_cannot_overlap_p): Take the
1873 constants as poly_widest_int rather than widest_int.
1874 (aff_combination_constant_multiple_p): Return the multiplier
1875 as a poly_widest_int.
1876 (aff_combination_zero_p, aff_combination_singleton_var_p): Handle
1878 * tree-affine.c (wide_int_ext_for_comb): Make original widest_int
1879 version static and add an overload for poly_widest_int.
1880 (aff_combination_const, aff_combination_add_cst)
1881 (wide_int_constant_multiple_p, aff_comb_cannot_overlap_p): Take
1882 the constants as poly_widest_int rather than widest_int.
1883 (tree_to_aff_combination): Generalize INTEGER_CST case to
1885 (aff_combination_to_tree): Track offsets as poly_widest_ints.
1886 (aff_combination_add_product, aff_combination_mult): Handle
1888 (aff_combination_constant_multiple_p): Return the multiplier
1889 as a poly_widest_int.
1890 * tree-predcom.c (determine_offset): Return the offset as a
1892 (split_data_refs_to_components, suitable_component_p): Update
1894 (valid_initializer_p): Update call to
1895 aff_combination_constant_multiple_p.
1896 * tree-ssa-address.c (addr_to_parts): Handle polynomial offsets.
1897 * tree-ssa-loop-ivopts.c (get_address_cost_ainc): Take the step
1898 as a poly_int64 rather than a HOST_WIDE_INT.
1899 (get_address_cost): Handle polynomial offsets.
1900 (iv_elimination_compare_lt): Likewise.
1901 (rewrite_use_nonlinear_expr): Likewise.
1903 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1904 Alan Hayward <alan.hayward@arm.com>
1905 David Sherwood <david.sherwood@arm.com>
1907 * tree-dfa.h (get_addr_base_and_unit_offset_1): Return the offset
1908 as a poly_int64_pod rather than a HOST_WIDE_INT.
1909 (get_addr_base_and_unit_offset): Likewise.
1910 * tree-dfa.c (get_addr_base_and_unit_offset_1): Likewise.
1911 (get_addr_base_and_unit_offset): Likewise.
1912 * doc/match-and-simplify.texi: Change off from HOST_WIDE_INT
1913 to poly_int64 in example.
1914 * fold-const.c (fold_binary_loc): Update call to
1915 get_addr_base_and_unit_offset.
1916 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
1917 (maybe_canonicalize_mem_ref_addr): Likewise.
1918 (gimple_fold_stmt_to_constant_1): Likewise.
1919 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
1921 * ipa-param-manipulation.c (ipa_modify_call_arguments): Likewise.
1922 * match.pd: Likewise.
1923 * omp-low.c (lower_omp_target): Likewise.
1924 * tree-sra.c (build_ref_for_offset): Likewise.
1925 (build_debug_ref_for_model): Likewise.
1926 * tree-ssa-address.c (maybe_fold_tmr): Likewise.
1927 * tree-ssa-alias.c (ao_ref_init_from_ptr_and_size): Likewise.
1928 * tree-ssa-ccp.c (optimize_memcpy): Likewise.
1929 * tree-ssa-forwprop.c (forward_propagate_addr_expr_1): Likewise.
1930 (constant_pointer_difference): Likewise.
1931 * tree-ssa-loop-niter.c (expand_simple_operations): Likewise.
1932 * tree-ssa-phiopt.c (jump_function_from_stmt): Likewise.
1933 * tree-ssa-pre.c (create_component_ref_by_pieces_1): Likewise.
1934 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
1935 (vn_reference_maybe_forwprop_address, vn_reference_lookup_3): Likewise.
1936 (set_ssa_val_to): Likewise.
1937 * tree-ssa-strlen.c (get_addr_stridx, addr_stridxptr)
1938 (maybe_diag_stxncpy_trunc): Likewise.
1939 * tree-vrp.c (vrp_prop::check_array_ref): Likewise.
1940 * tree.c (build_simple_mem_ref_loc): Likewise.
1941 (array_at_struct_end_p): Likewise.
1943 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1944 Alan Hayward <alan.hayward@arm.com>
1945 David Sherwood <david.sherwood@arm.com>
1947 * tree-dfa.h (get_ref_base_and_extent): Return the base, size and
1948 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1949 (get_ref_base_and_extent_hwi): Declare.
1950 * tree-dfa.c (get_ref_base_and_extent): Return the base, size and
1951 max_size as poly_int64_pods rather than HOST_WIDE_INTs.
1952 (get_ref_base_and_extent_hwi): New function.
1953 * cfgexpand.c (expand_debug_expr): Update call to
1954 get_ref_base_and_extent.
1955 * dwarf2out.c (add_var_loc_to_decl): Likewise.
1956 * gimple-fold.c (get_base_constructor): Return the offset as a
1957 poly_int64_pod rather than a HOST_WIDE_INT.
1958 (fold_const_aggregate_ref_1): Track polynomial sizes and offsets.
1959 * ipa-polymorphic-call.c
1960 (ipa_polymorphic_call_context::set_by_invariant)
1961 (extr_type_from_vtbl_ptr_store): Track polynomial offsets.
1962 (ipa_polymorphic_call_context::ipa_polymorphic_call_context)
1963 (check_stmt_for_type_change): Use get_ref_base_and_extent_hwi
1964 rather than get_ref_base_and_extent.
1965 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
1966 * ipa-prop.c (ipa_load_from_parm_agg, compute_complex_assign_jump_func)
1967 (get_ancestor_addr_info, determine_locally_known_aggregate_parts):
1969 * ipa-param-manipulation.c (ipa_get_adjustment_candidate): Update
1970 call to get_ref_base_and_extent.
1971 * tree-sra.c (create_access, get_access_for_expr): Likewise.
1972 * tree-ssa-alias.c (ao_ref_base, aliasing_component_refs_p)
1973 (stmt_kills_ref_p): Likewise.
1974 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
1975 * tree-ssa-scopedtables.c (avail_expr_hash, equal_mem_array_ref_p):
1977 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
1978 Use get_ref_base_and_extent_hwi rather than get_ref_base_and_extent
1979 when calling native_encode_expr.
1980 * tree-ssa-structalias.c (get_constraint_for_component_ref): Update
1981 call to get_ref_base_and_extent.
1982 (do_structure_copy): Use get_ref_base_and_extent_hwi rather than
1983 get_ref_base_and_extent.
1984 * var-tracking.c (track_expr_p): Likewise.
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 * ipa-param-manipulation.h (ipa_parm_adjustment::offset): Change from
1991 HOST_WIDE_INT to poly_int64_pod.
1992 * ipa-param-manipulation.c (ipa_modify_call_arguments): Track
1993 polynomail parameter offsets.
1995 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
1996 Alan Hayward <alan.hayward@arm.com>
1997 David Sherwood <david.sherwood@arm.com>
1999 * gengtype.c (main): Handle poly_int64_pod.
2000 * dwarf2out.h (dw_cfi_oprnd_cfa_loc): New dw_cfi_oprnd_type.
2001 (dw_cfi_oprnd::dw_cfi_cfa_loc): New field.
2002 (dw_cfa_location::offset, dw_cfa_location::base_offset): Change
2003 from HOST_WIDE_INT to poly_int64_pod.
2004 * dwarf2cfi.c (queued_reg_save::cfa_offset): Likewise.
2005 (copy_cfa): New function.
2006 (lookup_cfa_1): Use the cached dw_cfi_cfa_loc, if it exists.
2007 (cfi_oprnd_equal_p): Handle dw_cfi_oprnd_cfa_loc.
2008 (cfa_equal_p, dwarf2out_frame_debug_adjust_cfa)
2009 (dwarf2out_frame_debug_cfa_offset, dwarf2out_frame_debug_expr)
2010 (initial_return_save): Treat offsets as poly_ints.
2011 (def_cfa_0): Likewise. Cache the CFA in dw_cfi_cfa_loc if either
2012 offset is nonconstant.
2013 (reg_save): Take the offset as a poly_int64. Fall back to
2014 DW_CFA_expression for nonconstant offsets.
2015 (queue_reg_save): Take the offset as a poly_int64.
2016 * dwarf2out.c (dw_cfi_oprnd2_desc): Handle DW_CFA_def_cfa_expression.
2018 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2019 Alan Hayward <alan.hayward@arm.com>
2020 David Sherwood <david.sherwood@arm.com>
2022 * rtl.h (operand_subword, operand_subword_force): Take the offset
2023 as a poly_uint64 an unsigned int.
2024 * emit-rtl.c (operand_subword, operand_subword_force): Likewise.
2026 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2027 Alan Hayward <alan.hayward@arm.com>
2028 David Sherwood <david.sherwood@arm.com>
2030 * doc/rtl.texi: Update documentation of SUBREG_BYTE. Document the
2031 'p' format code. Use INT_LIST rather than SUBREG as the example of
2032 a code with an XINT and an XEXP. Remove the implication that
2033 accessing an rtx field using XINT is expected to work.
2034 * rtl.def (SUBREG): Change format from "ei" to "ep".
2035 * rtl.h (rtunion::rt_subreg): New field.
2036 (XCSUBREG): New macro.
2037 (SUBREG_BYTE): Use it.
2038 (subreg_shape): Change offset from an unsigned int to a poly_uint16.
2039 Update constructor accordingly.
2040 (subreg_shape::operator ==): Update accordingly.
2041 (subreg_shape::unique_id): Return an unsigned HOST_WIDE_INT rather
2042 than an unsigned int.
2043 (subreg_lsb, subreg_lowpart_offset, subreg_highpart_offset): Return
2044 a poly_uint64 rather than an unsigned int.
2045 (subreg_lsb_1): Likewise. Take the offset as a poly_uint64 rather
2046 than an unsigned int.
2047 (subreg_size_offset_from_lsb, subreg_size_lowpart_offset)
2048 (subreg_size_highpart_offset): Return a poly_uint64 rather than
2049 an unsigned int. Take the sizes as poly_uint64s.
2050 (subreg_offset_from_lsb): Return a poly_uint64 rather than
2051 an unsigned int. Take the shift as a poly_uint64 rather than
2053 (subreg_regno_offset, subreg_offset_representable_p): Take the offset
2054 as a poly_uint64 rather than an unsigned int.
2055 (simplify_subreg_regno): Likewise.
2056 (byte_lowpart_offset): Return the memory offset as a poly_int64
2058 (subreg_memory_offset): Likewise. Take the subreg offset as a
2059 poly_uint64 rather than an unsigned int.
2060 (simplify_subreg, simplify_gen_subreg, subreg_get_info)
2061 (gen_rtx_SUBREG, validate_subreg): Take the subreg offset as a
2062 poly_uint64 rather than an unsigned int.
2063 * rtl.c (rtx_format): Describe 'p' in comment.
2064 (copy_rtx, rtx_equal_p_cb, rtx_equal_p): Handle 'p'.
2065 * emit-rtl.c (validate_subreg, gen_rtx_SUBREG): Take the subreg
2066 offset as a poly_uint64 rather than an unsigned int.
2067 (byte_lowpart_offset): Return the memory offset as a poly_int64
2069 (subreg_memory_offset): Likewise. Take the subreg offset as a
2070 poly_uint64 rather than an unsigned int.
2071 (subreg_size_lowpart_offset, subreg_size_highpart_offset): Take the
2072 mode sizes as poly_uint64s rather than unsigned ints. Return a
2073 poly_uint64 rather than an unsigned int.
2074 (subreg_lowpart_p): Treat subreg offsets as poly_ints.
2075 (copy_insn_1): Handle 'p'.
2076 * rtlanal.c (set_noop_p): Treat subregs offsets as poly_uint64s.
2077 (subreg_lsb_1): Take the subreg offset as a poly_uint64 rather than
2078 an unsigned int. Return the shift in the same way.
2079 (subreg_lsb): Return the shift as a poly_uint64 rather than an
2081 (subreg_size_offset_from_lsb): Take the sizes and shift as
2082 poly_uint64s rather than unsigned ints. Return the offset as
2084 (subreg_get_info, subreg_regno_offset, subreg_offset_representable_p)
2085 (simplify_subreg_regno): Take the offset as a poly_uint64 rather than
2087 * rtlhash.c (add_rtx): Handle 'p'.
2088 * genemit.c (gen_exp): Likewise.
2089 * gengenrtl.c (type_from_format, gendef): Likewise.
2090 * gensupport.c (subst_pattern_match, get_alternatives_number)
2091 (collect_insn_data, alter_predicate_for_insn, alter_constraints)
2092 (subst_dup): Likewise.
2093 * gengtype.c (adjust_field_rtx_def): Likewise.
2094 * genrecog.c (find_operand, find_matching_operand, validate_pattern)
2095 (match_pattern_2): Likewise.
2096 (rtx_test::SUBREG_FIELD): New rtx_test::kind_enum.
2097 (rtx_test::subreg_field): New function.
2098 (operator ==, safe_to_hoist_p, transition_parameter_type)
2099 (print_nonbool_test, print_test): Handle SUBREG_FIELD.
2100 * genattrtab.c (attr_rtx_1): Say that 'p' is deliberately not handled.
2101 * genpeep.c (match_rtx): Likewise.
2102 * print-rtl.c (print_poly_int): Include if GENERATOR_FILE too.
2103 (rtx_writer::print_rtx_operand): Handle 'p'.
2104 (print_value): Handle SUBREG.
2105 * read-rtl.c (apply_int_iterator): Likewise.
2106 (rtx_reader::read_rtx_operand): Handle 'p'.
2107 * alias.c (rtx_equal_for_memref_p): Likewise.
2108 * cselib.c (rtx_equal_for_cselib_1, cselib_hash_rtx): Likewise.
2109 * caller-save.c (replace_reg_with_saved_mem): Treat subreg offsets
2111 * calls.c (expand_call): Likewise.
2112 * combine.c (combine_simplify_rtx, expand_field_assignment): Likewise.
2113 (make_extraction, gen_lowpart_for_combine): Likewise.
2114 * loop-invariant.c (hash_invariant_expr_1, invariant_expr_equal_p):
2116 * cse.c (remove_invalid_subreg_refs): Take the offset as a poly_uint64
2117 rather than an unsigned int. Treat subreg offsets as poly_ints.
2118 (exp_equiv_p): Handle 'p'.
2119 (hash_rtx_cb): Likewise. Treat subreg offsets as poly_ints.
2120 (equiv_constant, cse_insn): Treat subreg offsets as poly_ints.
2121 * dse.c (find_shift_sequence): Likewise.
2122 * dwarf2out.c (rtl_for_decl_location): Likewise.
2123 * expmed.c (extract_low_bits): Likewise.
2124 * expr.c (emit_group_store, undefined_operand_subword_p): Likewise.
2125 (expand_expr_real_2): Likewise.
2126 * final.c (alter_subreg): Likewise.
2127 (leaf_renumber_regs_insn): Handle 'p'.
2128 * function.c (assign_parm_find_stack_rtl, assign_parm_setup_stack):
2129 Treat subreg offsets as poly_ints.
2130 * fwprop.c (forward_propagate_and_simplify): Likewise.
2131 * ifcvt.c (noce_emit_move_insn, noce_emit_cmove): Likewise.
2132 * ira.c (get_subreg_tracking_sizes): Likewise.
2133 * ira-conflicts.c (go_through_subreg): Likewise.
2134 * ira-lives.c (process_single_reg_class_operands): Likewise.
2135 * jump.c (rtx_renumbered_equal_p): Likewise. Handle 'p'.
2136 * lower-subreg.c (simplify_subreg_concatn): Take the subreg offset
2137 as a poly_uint64 rather than an unsigned int.
2138 (simplify_gen_subreg_concatn, resolve_simple_move): Treat
2139 subreg offsets as poly_ints.
2140 * lra-constraints.c (operands_match_p): Handle 'p'.
2141 (match_reload, curr_insn_transform): Treat subreg offsets as poly_ints.
2142 * lra-spills.c (assign_mem_slot): Likewise.
2143 * postreload.c (move2add_valid_value_p): Likewise.
2144 * recog.c (general_operand, indirect_operand): Likewise.
2145 * regcprop.c (copy_value, maybe_mode_change): Likewise.
2146 (copyprop_hardreg_forward_1): Likewise.
2147 * reginfo.c (simplifiable_subregs_hasher::hash, simplifiable_subregs)
2148 (record_subregs_of_mode): Likewise.
2149 * rtlhooks.c (gen_lowpart_general, gen_lowpart_if_possible): Likewise.
2150 * reload.c (operands_match_p): Handle 'p'.
2151 (find_reloads_subreg_address): Treat subreg offsets as poly_ints.
2152 * reload1.c (alter_reg, choose_reload_regs): Likewise.
2153 (compute_reload_subreg_offset): Likewise, and return an poly_int64.
2154 * simplify-rtx.c (simplify_truncation, simplify_binary_operation_1):
2155 (test_vector_ops_duplicate): Treat subreg offsets as poly_ints.
2156 (simplify_const_poly_int_tests<N>::run): Likewise.
2157 (simplify_subreg, simplify_gen_subreg): Take the subreg offset as
2158 a poly_uint64 rather than an unsigned int.
2159 * valtrack.c (debug_lowpart_subreg): Likewise.
2160 * var-tracking.c (var_lowpart): Likewise.
2161 (loc_cmp): Handle 'p'.
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 * ira.c (get_subreg_tracking_sizes): New function.
2168 (init_live_subregs): Take an integer size rather than a register.
2169 (build_insn_chain): Use get_subreg_tracking_sizes. Update calls
2170 to init_live_subregs.
2172 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2173 Alan Hayward <alan.hayward@arm.com>
2174 David Sherwood <david.sherwood@arm.com>
2176 * expr.c (store_constructor_field): Change bitsize from a
2177 unsigned HOST_WIDE_INT to a poly_uint64 and bitpos from a
2178 HOST_WIDE_INT to a poly_int64.
2179 (store_constructor): Change size from a HOST_WIDE_INT to
2181 (store_field): Likewise bitsize and bitpos.
2183 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2184 Alan Hayward <alan.hayward@arm.com>
2185 David Sherwood <david.sherwood@arm.com>
2187 * expmed.h (store_bit_field): Change bitregion_start and
2188 bitregion_end from unsigned HOST_WIDE_INT to poly_uint64.
2189 * expmed.c (adjust_bit_field_mem_for_reg, strict_volatile_bitfield_p)
2190 (store_bit_field_1, store_integral_bit_field, store_bit_field)
2191 (store_fixed_bit_field, store_split_bit_field): Likewise.
2192 * expr.c (store_constructor_field, store_field): Likewise.
2193 (optimize_bitfield_assignment_op): Likewise. Make the same change
2194 to bitsize and bitpos.
2195 * machmode.h (bit_field_mode_iterator): Change m_bitregion_start
2196 and m_bitregion_end from HOST_WIDE_INT to poly_int64. Make the
2197 same change in the constructor arguments.
2198 (get_best_mode): Change bitregion_start and bitregion_end from
2199 unsigned HOST_WIDE_INT to poly_uint64.
2200 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2201 Change bitregion_start and bitregion_end from HOST_WIDE_INT to
2203 (bit_field_mode_iterator::next_mode): Update for new types
2204 of m_bitregion_start and m_bitregion_end.
2205 (get_best_mode): Change bitregion_start and bitregion_end from
2206 unsigned HOST_WIDE_INT to poly_uint64.
2208 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2209 Alan Hayward <alan.hayward@arm.com>
2210 David Sherwood <david.sherwood@arm.com>
2212 * rtl.h (simplify_gen_subreg): Add a temporary overload that
2213 accepts poly_uint64 offsets.
2214 * expmed.h (extract_bit_field): Take bitsize and bitnum as
2215 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2216 * expmed.c (lowpart_bit_field_p): Likewise.
2217 (extract_bit_field_as_subreg): New function, split out from...
2218 (extract_bit_field_1): ...here. Take bitsize and bitnum as
2219 poly_uint64s rather than unsigned HOST_WIDE_INTs. For vector
2220 extractions, check that BITSIZE matches the size of the extracted
2221 value and that BITNUM is an exact multiple of that size.
2222 If all else fails, try forcing the value into memory if
2223 BITNUM is variable, and adjusting the address so that the
2224 offset is constant. Split the part that can only handle constant
2225 bitsize and bitnum out into...
2226 (extract_integral_bit_field): ...this new function.
2227 (extract_bit_field): Take bitsize and bitnum as poly_uint64s
2228 rather than unsigned HOST_WIDE_INTs.
2230 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2231 Alan Hayward <alan.hayward@arm.com>
2232 David Sherwood <david.sherwood@arm.com>
2234 * expmed.h (store_bit_field): Take bitsize and bitnum as
2235 poly_uint64s rather than unsigned HOST_WIDE_INTs.
2236 * expmed.c (simple_mem_bitfield_p): Likewise. Add a parameter
2237 that returns the byte size.
2238 (store_bit_field_1): Take bitsize and bitnum as
2239 poly_uint64s rather than unsigned HOST_WIDE_INTs. Update call
2240 to simple_mem_bitfield_p. Split the part that can only handle
2241 constant bitsize and bitnum out into...
2242 (store_integral_bit_field): ...this new function.
2243 (store_bit_field): Take bitsize and bitnum as poly_uint64s rather
2244 than unsigned HOST_WIDE_INTs.
2245 (extract_bit_field_1): Update call to simple_mem_bitfield_p.
2247 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2248 Alan Hayward <alan.hayward@arm.com>
2249 David Sherwood <david.sherwood@arm.com>
2251 * lra-int.h (lra_reg): Change offset from int to poly_int64.
2252 (lra_insn_recog_data): Change sp_offset from HOST_WIDE_INT
2254 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2255 update_sp_offset from a HOST_WIDE_INT to a poly_int64.
2256 (lra_update_reg_val_offset, lra_reg_val_equal_p): Take the
2257 offset as a poly_int64 rather than an int.
2258 * lra-assigns.c (find_hard_regno_for_1): Handle poly_int64 offsets.
2259 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
2260 * lra-constraints.c (equiv_address_substitution): Track offsets
2262 (emit_inc): Check poly_int_rtx_p instead of CONST_INT_P.
2263 (curr_insn_transform): Handle the new form of sp_offset.
2264 * lra-eliminations.c (lra_elim_table): Change previous_offset
2265 and offset from HOST_WIDE_INT to poly_int64.
2266 (print_elim_table, update_reg_eliminate): Update accordingly.
2267 (self_elim_offsets): Change from HOST_WIDE_INT to poly_int64_pod.
2268 (get_elimination): Update accordingly.
2269 (form_sum): Check poly_int_rtx_p instead of CONST_INT_P.
2270 (lra_eliminate_regs_1, eliminate_regs_in_insn): Change
2271 update_sp_offset from a HOST_WIDE_INT to a poly_int64. Handle
2272 poly_int64 offsets generally.
2273 (curr_sp_change): Change from HOST_WIDE_INT to poly_int64.
2274 (mark_not_eliminable, init_elimination): Update accordingly.
2275 (remove_reg_equal_offset_note): Return a bool and pass the new
2276 offset back by pointer as a poly_int64.
2277 * lra-remat.c (change_sp_offset): Take sp_offset as a poly_int64
2278 rather than a HOST_WIDE_INT.
2279 (do_remat): Track offsets poly_int64s.
2280 * lra.c (lra_update_insn_recog_data, setup_sp_offset): Likewise.
2282 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2283 Alan Hayward <alan.hayward@arm.com>
2284 David Sherwood <david.sherwood@arm.com>
2286 * rtl.h (mem_attrs): Add a default constructor. Change size and
2287 offset from HOST_WIDE_INT to poly_int64.
2288 * emit-rtl.h (set_mem_offset, set_mem_size, adjust_address_1)
2289 (adjust_automodify_address_1, set_mem_attributes_minus_bitpos)
2290 (widen_memory_access): Take the sizes and offsets as poly_int64s
2291 rather than HOST_WIDE_INTs.
2292 * alias.c (ao_ref_from_mem): Handle the new form of MEM_OFFSET.
2293 (offset_overlap_p): Take poly_int64s rather than HOST_WIDE_INTs
2295 (adjust_offset_for_component_ref): Change the offset from a
2296 HOST_WIDE_INT to a poly_int64.
2297 (nonoverlapping_memrefs_p): Track polynomial offsets and sizes.
2298 * cfgcleanup.c (merge_memattrs): Update after mem_attrs changes.
2299 * dce.c (find_call_stack_args): Likewise.
2300 * dse.c (record_store): Likewise.
2301 * dwarf2out.c (tls_mem_loc_descriptor, dw_sra_loc_expr): Likewise.
2302 * print-rtl.c (rtx_writer::print_rtx): Likewise.
2303 * read-rtl-function.c (test_loading_mem): Likewise.
2304 * rtlanal.c (may_trap_p_1): Likewise.
2305 * simplify-rtx.c (delegitimize_mem_from_attrs): Likewise.
2306 * var-tracking.c (int_mem_offset, track_expr_p): Likewise.
2307 * emit-rtl.c (mem_attrs_eq_p, get_mem_align_offset): Likewise.
2308 (mem_attrs::mem_attrs): New function.
2309 (set_mem_attributes_minus_bitpos): Change bitpos from a
2310 HOST_WIDE_INT to poly_int64.
2311 (set_mem_alias_set, set_mem_addr_space, set_mem_align, set_mem_expr)
2312 (clear_mem_offset, clear_mem_size, change_address)
2313 (get_spill_slot_decl, set_mem_attrs_for_spill): Directly
2314 initialize mem_attrs.
2315 (set_mem_offset, set_mem_size, adjust_address_1)
2316 (adjust_automodify_address_1, offset_address, widen_memory_access):
2317 Likewise. Take poly_int64s rather than HOST_WIDE_INT.
2319 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2320 Alan Hayward <alan.hayward@arm.com>
2321 David Sherwood <david.sherwood@arm.com>
2323 * rtlanal.c (rtx_addr_can_trap_p_1): Take the offset and size
2324 as poly_int64s rather than HOST_WIDE_INTs. Use a size of -1
2325 rather than 0 to represent an unknown size. Assert that the size
2326 is known when the mode isn't BLKmode.
2327 (may_trap_p_1): Use -1 for unknown sizes.
2328 (rtx_addr_can_trap_p): Likewise. Pass BLKmode rather than VOIDmode.
2330 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2331 Alan Hayward <alan.hayward@arm.com>
2332 David Sherwood <david.sherwood@arm.com>
2334 * dse.c (store_info): Change offset and width from HOST_WIDE_INT
2335 to poly_int64. Update commentary for positions_needed.large.
2336 (read_info_type): Change offset and width from HOST_WIDE_INT
2338 (set_usage_bits): Likewise.
2339 (canon_address): Return the offset as a poly_int64 rather than
2340 a HOST_WIDE_INT. Use strip_offset_and_add.
2341 (set_all_positions_unneeded, any_positions_needed_p): Use
2342 positions_needed.large to track stores with non-constant widths.
2343 (all_positions_needed_p): Likewise. Take the offset and width
2344 as poly_int64s rather than ints. Assert that rhs is nonnull.
2345 (record_store): Cope with non-constant offsets and widths.
2346 Nullify the rhs of an earlier store if we can't tell which bytes
2348 (find_shift_sequence): Take the access_size and shift as poly_int64s
2350 (get_stored_val): Take the read_offset and read_width as poly_int64s
2351 rather than HOST_WIDE_INTs.
2352 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Handle
2353 non-constant offsets and widths.
2355 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2356 Alan Hayward <alan.hayward@arm.com>
2357 David Sherwood <david.sherwood@arm.com>
2359 * inchash.h (inchash::hash::add_poly_int): New function.
2360 * tree-ssa-alias.h (ao_ref::offset, ao_ref::size, ao_ref::max_size):
2361 Use poly_int64 rather than HOST_WIDE_INT.
2362 (ao_ref::max_size_known_p): New function.
2363 * tree-ssa-sccvn.h (vn_reference_op_struct::off): Use poly_int64_pod
2364 rather than HOST_WIDE_INT.
2365 * tree-ssa-alias.c (ao_ref_base): Apply get_ref_base_and_extent
2366 to temporaries until its interface is adjusted to match.
2367 (ao_ref_init_from_ptr_and_size): Handle polynomial offsets and sizes.
2368 (aliasing_component_refs_p, decl_refs_may_alias_p)
2369 (indirect_ref_may_alias_decl_p, indirect_refs_may_alias_p): Take
2370 the offsets and max_sizes as poly_int64s instead of HOST_WIDE_INTs.
2371 (refs_may_alias_p_1, stmt_kills_ref_p): Adjust for changes to
2373 * alias.c (ao_ref_from_mem): Likewise.
2374 * tree-ssa-dce.c (mark_aliased_reaching_defs_necessary_1): Likewise.
2375 * tree-ssa-dse.c (valid_ao_ref_for_dse, normalize_ref)
2376 (clear_bytes_written_by, setup_live_bytes_from_ref, compute_trims)
2377 (maybe_trim_complex_store, maybe_trim_constructor_store)
2378 (live_bytes_read, dse_classify_store): Likewise.
2379 * tree-ssa-sccvn.c (vn_reference_compute_hash, vn_reference_eq):
2380 (copy_reference_ops_from_ref, ao_ref_init_from_vn_reference)
2381 (fully_constant_vn_reference_p, valueize_refs_1): Likewise.
2382 (vn_reference_lookup_3): Likewise.
2383 * tree-ssa-uninit.c (warn_uninitialized_vars): Likewise.
2385 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2386 Alan Hayward <alan.hayward@arm.com>
2387 David Sherwood <david.sherwood@arm.com>
2389 * tree-ssa-alias.c (indirect_ref_may_alias_decl_p)
2390 (indirect_refs_may_alias_p): Use ranges_may_overlap_p
2391 instead of ranges_overlap_p.
2393 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2394 Alan Hayward <alan.hayward@arm.com>
2395 David Sherwood <david.sherwood@arm.com>
2397 * tree-ssa-alias.c (same_addr_size_stores_p): Take the offsets and
2398 sizes as poly_int64s rather than HOST_WIDE_INTs.
2400 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2401 Alan Hayward <alan.hayward@arm.com>
2402 David Sherwood <david.sherwood@arm.com>
2404 * gimple-fold.h (fold_ctor_reference): Take the offset and size
2405 as poly_uint64 rather than unsigned HOST_WIDE_INT.
2406 * gimple-fold.c (fold_ctor_reference): Likewise.
2408 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2409 Alan Hayward <alan.hayward@arm.com>
2410 David Sherwood <david.sherwood@arm.com>
2412 * target.def (dwarf_poly_indeterminate_value): New hook.
2413 * targhooks.h (default_dwarf_poly_indeterminate_value): Declare.
2414 * targhooks.c (default_dwarf_poly_indeterminate_value): New function.
2415 * doc/tm.texi.in (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Document.
2416 * doc/tm.texi: Regenerate.
2417 * dwarf2out.h (build_cfa_loc, build_cfa_aligned_loc): Take the
2418 offset as a poly_int64.
2419 * dwarf2out.c (new_reg_loc_descr): Move later in file. Take the
2420 offset as a poly_int64.
2421 (loc_descr_plus_const, loc_list_plus_const, build_cfa_aligned_loc):
2422 Take the offset as a poly_int64.
2423 (build_cfa_loc): Likewise. Use loc_descr_plus_const.
2424 (frame_pointer_fb_offset): Change to a poly_int64.
2425 (int_loc_descriptor): Take the offset as a poly_int64. Use
2426 targetm.dwarf_poly_indeterminate_value for polynomial offsets.
2427 (based_loc_descr): Take the offset as a poly_int64.
2428 Use strip_offset_and_add to handle (plus X (const)).
2429 Use new_reg_loc_descr instead of an open-coded version of the
2430 previous implementation.
2431 (mem_loc_descriptor): Handle CONST_POLY_INT.
2432 (compute_frame_pointer_to_fb_displacement): Take the offset as a
2433 poly_int64. Use strip_offset_and_add to handle (plus X (const)).
2435 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2436 Alan Hayward <alan.hayward@arm.com>
2437 David Sherwood <david.sherwood@arm.com>
2439 * rtl.h (reg_attrs::offset): Change from HOST_WIDE_INT to poly_int64.
2440 (gen_rtx_REG_offset): Take the offset as a poly_int64.
2441 * inchash.h (inchash::hash::add_poly_hwi): New function.
2442 * gengtype.c (main): Register poly_int64.
2443 * emit-rtl.c (reg_attr_hasher::hash): Use inchash. Treat the
2444 offset as a poly_int.
2445 (reg_attr_hasher::equal): Use must_eq to compare offsets.
2446 (get_reg_attrs, update_reg_offset, gen_rtx_REG_offset): Take the
2447 offset as a poly_int64.
2448 (set_reg_attrs_from_value): Treat the offset as a poly_int64.
2449 * print-rtl.c (print_poly_int): New function.
2450 (rtx_writer::print_rtx_operand_code_r): Treat REG_OFFSET as
2452 * var-tracking.c (track_offset_p, get_tracked_reg_offset): New
2454 (var_reg_set, var_reg_delete_and_set, var_reg_delete): Use them.
2455 (same_variable_part_p, track_loc_p): Take the offset as a poly_int64.
2456 (vt_get_decl_and_offset): Return the offset as a poly_int64.
2457 Enforce track_offset_p for parts of a PARALLEL.
2458 (vt_add_function_parameter): Use const_offset for the final
2459 offset to track. Use get_tracked_reg_offset for the parts
2462 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2463 Alan Hayward <alan.hayward@arm.com>
2464 David Sherwood <david.sherwood@arm.com>
2466 * target.def (truly_noop_truncation): Take poly_uint64s instead of
2467 unsigned ints. Change default to hook_bool_puint64_puint64_true.
2468 * doc/tm.texi: Regenerate.
2469 * hooks.h (hook_bool_uint_uint_true): Delete.
2470 (hook_bool_puint64_puint64_true): Declare.
2471 * hooks.c (hook_bool_uint_uint_true): Delete.
2472 (hook_bool_puint64_puint64_true): New function.
2473 * config/mips/mips.c (mips_truly_noop_truncation): Take poly_uint64s
2474 instead of unsigned ints.
2475 * config/spu/spu.c (spu_truly_noop_truncation): Likewise.
2476 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): Likewise.
2478 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2479 Alan Hayward <alan.hayward@arm.com>
2480 David Sherwood <david.sherwood@arm.com>
2482 * optabs.h (expand_operand): Add an int_value field.
2483 (create_expand_operand): Add an int_value parameter and use it
2484 to initialize the new expand_operand field.
2485 (create_integer_operand): Replace with a declaration of a function
2486 that accepts poly_int64s. Move the implementation to...
2487 * optabs.c (create_integer_operand): ...here.
2488 (maybe_legitimize_operand): For EXPAND_INTEGER, check whether
2489 the mode preserves the value of int_value, instead of calling
2490 const_int_operand on the rtx. Use gen_int_mode to generate
2493 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2494 Alan Hayward <alan.hayward@arm.com>
2495 David Sherwood <david.sherwood@arm.com>
2497 * dumpfile.h (dump_dec): Declare.
2498 * dumpfile.c (dump_dec): New function.
2499 * pretty-print.h (pp_wide_integer): Turn into a function and
2500 declare a poly_int version.
2501 * pretty-print.c (pp_wide_integer): New function for poly_ints.
2503 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2504 Alan Hayward <alan.hayward@arm.com>
2505 David Sherwood <david.sherwood@arm.com>
2507 * doc/generic.texi (POLY_INT_CST): Document.
2508 * tree.def (POLY_INT_CST): New tree code.
2509 * treestruct.def (TS_POLY_INT_CST): New tree layout.
2510 * tree-core.h (tree_poly_int_cst): New struct.
2511 (tree_node): Add a poly_int_cst field.
2512 * tree.h (POLY_INT_CST_P, POLY_INT_CST_COEFF): New macros.
2513 (wide_int_to_tree, force_fit_type): Take a poly_wide_int_ref
2514 instead of a wide_int_ref.
2515 (build_int_cst, build_int_cst_type): Take a poly_int64 instead
2517 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2518 instead of an unsigned HOST_WIDE_INT.
2519 (build_poly_int_cst, tree_fits_poly_int64_p, tree_fits_poly_uint64_p)
2520 (ptrdiff_tree_p): Declare.
2521 (tree_to_poly_int64, tree_to_poly_uint64): Likewise. Provide
2522 extern inline implementations if the target doesn't use POLY_INT_CST.
2523 (poly_int_tree_p): New function.
2524 (wi::unextended_tree): New class.
2525 (wi::int_traits <unextended_tree>): New override.
2526 (wi::extended_tree): Add a default constructor.
2527 (wi::extended_tree::get_tree): New function.
2528 (wi::widest_extended_tree, wi::offset_extended_tree): New typedefs.
2529 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): Use them.
2530 (wi::tree_to_poly_widest_ref, wi::tree_to_poly_offset_ref)
2531 (wi::tree_to_poly_wide_ref): New typedefs.
2532 (wi::ints_for): Provide overloads for extended_tree and
2534 (poly_int_cst_value, wi::to_poly_widest, wi::to_poly_offset)
2535 (wi::to_wide): New functions.
2536 (wi::fits_to_boolean_p, wi::fits_to_tree_p): Handle poly_ints.
2537 * tree.c (poly_int_cst_hasher): New struct.
2538 (poly_int_cst_hash_table): New variable.
2539 (tree_node_structure_for_code, tree_code_size, simple_cst_equal)
2540 (valid_constant_size_p, add_expr, drop_tree_overflow): Handle
2542 (initialize_tree_contains_struct): Handle TS_POLY_INT_CST.
2543 (init_ttree): Initialize poly_int_cst_hash_table.
2544 (build_int_cst, build_int_cst_type, build_invariant_address): Take
2545 a poly_int64 instead of a HOST_WIDE_INT.
2546 (build_int_cstu, build_array_type_nelts): Take a poly_uint64
2547 instead of an unsigned HOST_WIDE_INT.
2548 (wide_int_to_tree): Rename to...
2549 (wide_int_to_tree_1): ...this.
2550 (build_new_poly_int_cst, build_poly_int_cst): New functions.
2551 (force_fit_type): Take a poly_wide_int_ref instead of a wide_int_ref.
2552 (wide_int_to_tree): New function that takes a poly_wide_int_ref.
2553 (ptrdiff_tree_p, tree_to_poly_int64, tree_to_poly_uint64)
2554 (tree_fits_poly_int64_p, tree_fits_poly_uint64_p): New functions.
2555 * lto-streamer-out.c (DFS::DFS_write_tree_body, hash_tree): Handle
2557 * tree-streamer-in.c (lto_input_ts_poly_tree_pointers): Likewise.
2558 (streamer_read_tree_body): Likewise.
2559 * tree-streamer-out.c (write_ts_poly_tree_pointers): Likewise.
2560 (streamer_write_tree_body): Likewise.
2561 * tree-streamer.c (streamer_check_handled_ts_structures): Likewise.
2562 * asan.c (asan_protect_global): Require the size to be an INTEGER_CST.
2563 * cfgexpand.c (expand_debug_expr): Handle POLY_INT_CST.
2564 * expr.c (expand_expr_real_1, const_vector_from_tree): Likewise.
2565 * gimple-expr.h (is_gimple_constant): Likewise.
2566 * gimplify.c (maybe_with_size_expr): Likewise.
2567 * print-tree.c (print_node): Likewise.
2568 * tree-data-ref.c (data_ref_compare_tree): Likewise.
2569 * tree-pretty-print.c (dump_generic_node): Likewise.
2570 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
2571 * tree-vect-data-refs.c (dr_group_sort_cmp): Likewise.
2572 * tree-vrp.c (compare_values_warnv): Likewise.
2573 * tree-ssa-loop-ivopts.c (determine_base_object, constant_multiple_of)
2574 (get_loop_invariant_expr, add_candidate_1, get_computation_aff_1)
2575 (force_expr_to_var_cost): Likewise.
2576 * tree-ssa-loop.c (for_each_index): Likewise.
2577 * fold-const.h (build_invariant_address, size_int_kind): Take a
2578 poly_int64 instead of a HOST_WIDE_INT.
2579 * fold-const.c (fold_negate_expr_1, const_binop, const_unop)
2580 (fold_convert_const, multiple_of_p, fold_negate_const): Handle
2582 (size_binop_loc): Likewise. Allow int_const_binop_1 to fail.
2583 (int_const_binop_2): New function, split out from...
2584 (int_const_binop_1): ...here. Handle POLY_INT_CST.
2585 (size_int_kind): Take a poly_int64 instead of a HOST_WIDE_INT.
2586 * expmed.c (make_tree): Handle CONST_POLY_INT_P.
2587 * gimple-ssa-strength-reduction.c (slsr_process_add)
2588 (slsr_process_mul): Check for INTEGER_CSTs before using them
2590 * stor-layout.c (bits_from_bytes): New function.
2591 (bit_from_pos): Use it.
2592 (layout_type): Likewise. For vectors, multiply the TYPE_SIZE_UNIT
2593 by BITS_PER_UNIT to get the TYPE_SIZE.
2594 * tree-cfg.c (verify_expr, verify_types_in_gimple_reference): Allow
2595 MEM_REF and TARGET_MEM_REF offsets to be a POLY_INT_CST.
2597 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2598 Alan Hayward <alan.hayward@arm.com>
2599 David Sherwood <david.sherwood@arm.com>
2601 * doc/rtl.texi (const_poly_int): Document. Also document the
2602 rtl sharing behavior.
2603 * gengenrtl.c (excluded_rtx): Return true for CONST_POLY_INT.
2604 * rtl.h (const_poly_int_def): New struct.
2605 (rtx_def::u): Add a cpi field.
2606 (CASE_CONST_UNIQUE, CASE_CONST_ANY): Add CONST_POLY_INT.
2607 (CONST_POLY_INT_P, CONST_POLY_INT_COEFFS): New macros.
2608 (wi::rtx_to_poly_wide_ref): New typedef
2609 (const_poly_int_value, wi::to_poly_wide, rtx_to_poly_int64)
2610 (poly_int_rtx_p): New functions.
2611 (trunc_int_for_mode): Declare a poly_int64 version.
2612 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2613 (immed_wide_int_const): Take a poly_wide_int_ref rather than
2615 (strip_offset): Declare.
2616 (strip_offset_and_add): New function.
2617 * rtl.def (CONST_POLY_INT): New rtx code.
2618 * rtl.c (rtx_size): Handle CONST_POLY_INT.
2619 (shared_const_p): Use poly_int_rtx_p.
2620 * emit-rtl.h (gen_int_mode): Take a poly_int64 instead of a
2622 (gen_int_shift_amount): Likewise.
2623 * emit-rtl.c (const_poly_int_hasher): New class.
2624 (const_poly_int_htab): New variable.
2625 (init_emit_once): Initialize it when NUM_POLY_INT_COEFFS > 1.
2626 (const_poly_int_hasher::hash): New function.
2627 (const_poly_int_hasher::equal): Likewise.
2628 (gen_int_mode): Take a poly_int64 instead of a HOST_WIDE_INT.
2629 (immed_wide_int_const): Rename to...
2630 (immed_wide_int_const_1): ...this and make static.
2631 (immed_wide_int_const): New function, taking a poly_wide_int_ref
2632 instead of a wide_int_ref.
2633 (gen_int_shift_amount): Take a poly_int64 instead of a HOST_WIDE_INT.
2634 (gen_lowpart_common): Handle CONST_POLY_INT.
2635 * cse.c (hash_rtx_cb, equiv_constant): Likewise.
2636 * cselib.c (cselib_hash_rtx): Likewise.
2637 * dwarf2out.c (const_ok_for_output_1): Likewise.
2638 * expr.c (convert_modes): Likewise.
2639 * print-rtl.c (rtx_writer::print_rtx, print_value): Likewise.
2640 * rtlhash.c (add_rtx): Likewise.
2641 * explow.c (trunc_int_for_mode): Add a poly_int64 version.
2642 (plus_constant): Take a poly_int64 instead of a HOST_WIDE_INT.
2643 Handle existing CONST_POLY_INT rtxes.
2644 * expmed.h (expand_shift): Take a poly_int64 instead of a
2646 * expmed.c (expand_shift): Likewise.
2647 * rtlanal.c (strip_offset): New function.
2648 (commutative_operand_precedence): Give CONST_POLY_INT the same
2649 precedence as CONST_DOUBLE and put CONST_WIDE_INT between that
2651 * rtl-tests.c (const_poly_int_tests): New struct.
2652 (rtl_tests_c_tests): Use it.
2653 * simplify-rtx.c (simplify_const_unary_operation): Handle
2655 (simplify_const_binary_operation): Likewise.
2656 (simplify_binary_operation_1): Fold additions of symbolic constants
2657 and CONST_POLY_INTs.
2658 (simplify_subreg): Handle extensions and truncations of
2660 (simplify_const_poly_int_tests): New struct.
2661 (simplify_rtx_c_tests): Use it.
2662 * wide-int.h (storage_ref): Add default constructor.
2663 (wide_int_ref_storage): Likewise.
2664 (trailing_wide_ints): Use GTY((user)).
2665 (trailing_wide_ints::operator[]): Add a const version.
2666 (trailing_wide_ints::get_precision): New function.
2667 (trailing_wide_ints::extra_size): Likewise.
2669 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2670 Alan Hayward <alan.hayward@arm.com>
2671 David Sherwood <david.sherwood@arm.com>
2673 * emit-rtl.h (gen_int_shift_amount): Declare.
2674 * emit-rtl.c (gen_int_shift_amount): New function.
2675 * asan.c (asan_emit_stack_protection): Use gen_int_shift_amount
2677 * calls.c (shift_return_value): Likewise.
2678 * cse.c (fold_rtx): Likewise.
2679 * dse.c (find_shift_sequence): Likewise.
2680 * expmed.c (init_expmed_one_mode, store_bit_field_1, expand_shift_1)
2681 (expand_shift, expand_smod_pow2): Likewise.
2682 * lower-subreg.c (shift_cost): Likewise.
2683 * optabs.c (expand_superword_shift, expand_doubleword_mult)
2684 (expand_unop, expand_binop, shift_amt_for_vec_perm_mask)
2685 (expand_vec_perm_var): Likewise.
2686 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2687 (simplify_binary_operation_1): Likewise.
2688 * combine.c (try_combine, find_split_point, force_int_to_mode)
2689 (simplify_shift_const_1, simplify_shift_const): Likewise.
2690 (change_zero_ext): Likewise. Use simplify_gen_binary.
2692 2017-12-20 Richard Sandiford <richard.sandiford@linaro.org>
2694 * poly-int.h (multiple_p): Fix handling of two non-poly_ints.
2696 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2698 * doc/invoke.texi (ARM Options): Document accepted extension options
2699 for -march=armv8.3-a.
2701 2017-12-20 Richard Earnshaw <rearnsha@arm.com>
2704 * config.gcc (arm*-*-linux*): When configured with --with-float=hard
2705 or --with-float=softfp, set the default CPU to arm10e.
2707 2017-12-20 Eric Botcazou <ebotcazou@adacore.com>
2709 * config/visium/constraints.md (J, K, L): Use IN_RANGE macro.
2710 * config/visium/predicates.md (const_shift_operand): Likewise.
2711 * config/visium/visium.c (visium_legitimize_address): Fix oversight.
2712 (visium_legitimize_reload_address): Likewise.
2714 2017-12-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2717 * config/arm/arm.h (TEST_REGNO): Adjust comment as expected in r255830.
2719 2017-12-20 Jakub Jelinek <jakub@redhat.com>
2722 * calls.c (compute_argument_addresses): Ignore TYPE_EMPTY_P arguments.
2724 2017-12-20 Julia Koval <julia.koval@intel.com>
2726 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VPCLMULQDQ_SET,
2727 OPTION_MASK_ISA_VPCLMULQDQ_UNSET): New.
2728 (ix86_handle_option): Handle -mvpclmulqdq, move cx6 to flags2.
2729 * config.gcc: Include vpclmulqdqintrin.h.
2730 * config/i386/cpuid.h: Handle bit_VPCLMULQDQ.
2731 * config/i386/driver-i386.c (host_detect_local_cpu): Handle
2733 * config/i386/i386-builtin.def (__builtin_ia32_vpclmulqdq_v2di,
2734 __builtin_ia32_vpclmulqdq_v4di, __builtin_ia32_vpclmulqdq_v8di): New.
2735 * config/i386/i386-c.c (__VPCLMULQDQ__): New.
2736 * config/i386/i386.c (isa2_opts): Add -mcx16.
2737 (isa_opts): Add -mpclmulqdq, remove -mcx16.
2738 (ix86_option_override_internal): Move mcx16 to flags2.
2739 (ix86_valid_target_attribute_inner_p): Add vpclmulqdq.
2740 (ix86_expand_builtin): Handle OPTION_MASK_ISA_VPCLMULQDQ.
2741 * config/i386/i386.h (TARGET_VPCLMULQDQ, TARGET_VPCLMULQDQ_P): New.
2742 * config/i386/i386.opt: Add mvpclmulqdq, move mcx16 to flags2.
2743 * config/i386/immintrin.h: Include vpclmulqdqintrin.h.
2744 * config/i386/sse.md (vpclmulqdq_<mode>): New pattern.
2745 * config/i386/vpclmulqdqintrin.h (_mm512_clmulepi64_epi128,
2746 _mm_clmulepi64_epi128, _mm256_clmulepi64_epi128): New intrinsics.
2747 * doc/invoke.texi: Add -mvpclmulqdq.
2749 2017-12-20 Tom de Vries <tom@codesourcery.com>
2752 * config/i386/i386.c (ix86_static_chain): Move
2753 DECL_STATIC_CHAIN test ...
2754 * calls.c (rtx_for_static_chain): ... here. New function.
2755 * calls.h (rtx_for_static_chain): Declare.
2756 * builtins.c (expand_builtin_setjmp_receiver): Use rtx_for_static_chain
2757 instead of targetm.calls.static_chain.
2758 * df-scan.c (df_get_entry_block_def_set): Same.
2760 2017-12-19 Tom de Vries <tom@codesourcery.com>
2762 PR tree-optimization/83493
2763 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for): Unshare ub
2766 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2768 * gimple-ssa-sprintf.c (format_directive): Use inform_n instead of
2769 inform with hardcoded english plural handling.
2771 2017-12-18 Jeff Law <law@redhat.com>
2773 PR tree-optimization/83477
2774 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis): For
2775 a non-virtual PHI, always push a new range.
2777 2017-12-19 Martin Sebor <msebor@redhat.com>
2780 * builtins.c (compute_objsize): Handle non-constant offsets.
2782 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2784 PR tree-optimization/83444
2785 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): For the
2786 character load case, if get_stridx on MEM_REF's operand doesn't
2787 look usable, retry with get_addr_stridx.
2789 2017-12-19 Alexandre Oliva <aoliva@redhat.com>
2792 * var-tracking.c (vt_debug_insns_local): Do not drop markers.
2793 (variable_tracking_main_1): Keep markers even when VTA fails.
2796 * cfgexpand.c (expand_gimple_basic_block): Expand label first,
2797 even if there are markers before it.
2798 * cfgrtl.c (rtl_verify_bb_layout): Reject DEBUG_INSNs outside BBs.
2800 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2802 * read-rtl.c (parse_reg_note_name): Replace Yoda conditions with
2803 typical order conditions.
2804 * sel-sched.c (extract_new_fences_from): Likewise.
2805 * config/visium/constraints.md (J, K, L): Likewise.
2806 * config/visium/predicates.md (const_shift_operand): Likewise.
2807 * config/visium/visium.c (visium_legitimize_address,
2808 visium_legitimize_reload_address): Likewise.
2809 * config/m68k/m68k.c (output_reg_adjust, emit_reg_adjust): Likewise.
2810 * config/arm/arm.c (arm_block_move_unaligned_straight): Likewise.
2811 * config/avr/constraints.md (Y01, Ym1, Y02, Ym2): Likewise.
2812 * config/avr/avr-log.c (avr_vdump, avr_log_set_avr_log,
2813 SET_DUMP_DETAIL): Likewise.
2814 * config/avr/predicates.md (const_8_16_24_operand): Likewise.
2815 * config/avr/avr.c (STR_PREFIX_P, avr_popcount_each_byte,
2816 avr_is_casesi_sequence, avr_casei_sequence_check_operands,
2817 avr_set_core_architecture, avr_set_current_function,
2818 avr_legitimize_reload_address, avr_asm_len, avr_print_operand,
2819 output_movqi, output_movsisf, avr_out_plus, avr_out_bitop,
2820 avr_out_fract, avr_adjust_insn_length, avr_encode_section_info,
2821 avr_2word_insn_p, output_reload_in_const, avr_has_nibble_0xf,
2822 avr_map_decompose, avr_fold_builtin): Likewise.
2823 * config/avr/driver-avr.c (avr_devicespecs_file): Likewise.
2824 * config/avr/gen-avr-mmcu-specs.c (str_prefix_p, print_mcu): Likewise.
2825 * config/i386/i386.c (ix86_parse_stringop_strategy_string): Likewise.
2826 * config/m32c/m32c-pragma.c (m32c_pragma_memregs): Likewise.
2827 * config/m32c/m32c.c (m32c_conditional_register_usage,
2828 m32c_address_cost): Likewise.
2829 * config/m32c/predicates.md (shiftcount_operand,
2830 longshiftcount_operand): Likewise.
2831 * config/iq2000/iq2000.c (iq2000_expand_prologue): Likewise.
2832 * config/nios2/nios2.c (nios2_handle_custom_fpu_insn_option,
2833 can_use_cdx_ldstw): Likewise.
2834 * config/nios2/nios2.h (CDX_REG_P): Likewise.
2835 * config/cr16/cr16.h (RETURN_ADDR_RTX, REGNO_MODE_OK_FOR_BASE_P):
2837 * config/cr16/cr16.md (*mov<mode>_double): Likewise.
2838 * config/cr16/cr16.c (cr16_create_dwarf_for_multi_push): Likewise.
2839 * config/h8300/h8300.c (h8300_rtx_costs, get_shift_alg): Likewise.
2840 * config/vax/constraints.md (U06, U08, U16, CN6, S08, S16): Likewise.
2841 * config/vax/vax.c (adjacent_operands_p): Likewise.
2842 * config/ft32/constraints.md (L, b, KA): Likewise.
2843 * config/ft32/ft32.c (ft32_load_immediate, ft32_expand_prologue):
2845 * cfgexpand.c (expand_stack_alignment): Likewise.
2846 * gcse.c (insert_expr_in_table): Likewise.
2847 * print-rtl.c (rtx_writer::print_rtx_operand_codes_E_and_V): Likewise.
2848 * cgraphunit.c (cgraph_node::expand): Likewise.
2849 * ira-build.c (setup_min_max_allocno_live_range_point): Likewise.
2850 * emit-rtl.c (add_insn): Likewise.
2851 * input.c (dump_location_info): Likewise.
2852 * passes.c (NEXT_PASS): Likewise.
2853 * read-rtl-function.c (parse_note_insn_name,
2854 function_reader::read_rtx_operand_r, function_reader::parse_mem_expr):
2856 * sched-rgn.c (sched_rgn_init): Likewise.
2857 * diagnostic-show-locus.c (layout::show_ruler): Likewise.
2858 * combine.c (find_split_point, simplify_if_then_else, force_to_mode,
2859 if_then_else_cond, simplify_shift_const_1, simplify_comparison):
2861 * explow.c (eliminate_constant_term): Likewise.
2862 * final.c (leaf_renumber_regs_insn): Likewise.
2863 * cfgrtl.c (print_rtl_with_bb): Likewise.
2864 * genhooks.c (emit_init_macros): Likewise.
2865 * poly-int.h (maybe_ne, maybe_le, maybe_lt): Likewise.
2866 * tree-data-ref.c (conflict_fn): Likewise.
2867 * selftest.c (assert_streq): Likewise.
2868 * expr.c (store_constructor_field, expand_expr_real_1): Likewise.
2869 * fold-const.c (fold_range_test, extract_muldiv_1, fold_truth_andor,
2870 fold_binary_loc, multiple_of_p): Likewise.
2871 * reload.c (push_reload, find_equiv_reg): Likewise.
2872 * et-forest.c (et_nca, et_below): Likewise.
2873 * dbxout.c (dbxout_symbol_location): Likewise.
2874 * reorg.c (relax_delay_slots): Likewise.
2875 * dojump.c (do_compare_rtx_and_jump): Likewise.
2876 * gengtype-parse.c (type): Likewise.
2877 * simplify-rtx.c (simplify_gen_ternary, simplify_gen_relational,
2878 simplify_const_relational_operation): Likewise.
2879 * reload1.c (do_output_reload): Likewise.
2880 * dumpfile.c (get_dump_file_info_by_switch): Likewise.
2881 * gengtype.c (type_for_name): Likewise.
2882 * gimple-ssa-sprintf.c (format_directive): Likewise.
2884 2017-12-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2887 * config/arm/arm.h (TEST_REGNO): Check reg_renumber is set before
2888 accessing it. Adjust comment.
2890 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2893 * predict.c (zero_one_minusone): New function.
2894 (apply_return_prediction): Avoid return prediction for functions
2895 returning only -1, 0 and 1 values, unless they only return -1 and 0
2898 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2900 * config/arc/arc.c (legitimate_scaled_address_p): Clean
2901 fall-through warning.
2902 (arc_compute_frame_size): Remove unused variables.
2903 (arc_print_operand): Fix fprintif format.
2904 (arc_can_follow_jump): Clean fall-through warning.
2906 2017-12-19 Marek Polacek <polacek@redhat.com>
2909 * config/i386/i386.c (init_cumulative_args): Don't check TYPE_EMPTY_P
2912 2017-12-19 Claudiu Zissulescu <claziss@synopsys.com>
2914 * config/arc/arc.c (overriderregs): New variable.
2915 (arc_override_options): Track fixed/call saved/call options.
2916 (arc_conditional_register_usage): Check against overrideregs
2917 variable whenever we change register properties.
2919 2017-12-19 Nathan Sidwell <nathan@acm.org>
2921 * opts.c (finish_options): Don't prefix dump_base_name if it
2922 already contains directories.
2924 2017-12-19 Martin Liska <mliska@suse.cz>
2926 PR rtl-optimization/82675
2927 * loop-unroll.c (unroll_loop_constant_iterations): Allocate one
2928 more element in sbitmap.
2930 2017-12-19 Paolo Carlini <paolo.carlini@oracle.com>
2932 * gimplify.c (gimplify_expr): Use error_operand_p.
2934 2017-12-19 Sebastian Huber <sebastian.huber@embedded-brains.de>
2937 * config/rs6000/t-rtems (MULTILIB_REQUIRED): Remove 64-bit soft-float
2940 2017-12-19 Daniel Cederman <cederman@gaisler.com>
2942 * config/sparc/sparc.c (sparc_do_work_around_errata): Make sure
2943 the jump is to a label.
2945 2017-12-19 Jakub Jelinek <jakub@redhat.com>
2947 PR tree-optimization/83444
2948 * tree-ssa-strlen.c (strlen_check_and_optimize_stmt): Optimize
2953 * ipa-inline.c (flatten_remove_node_hook): New function.
2954 (ipa_inline): Keep only nodes with flatten attribute at the end of
2955 the array in the order from ipa_reverse_postorder, only walk that
2956 portion of array for flattening, if there is more than one such
2957 node, temporarily register a removal hook and ignore removed nodes.
2959 PR tree-optimization/80631
2960 * tree-vect-loop.c (vect_create_epilog_for_reduction): Compare
2961 induc_code against MAX_EXPR or MIN_EXPR instead of reduc_fn against
2962 IFN_REDUC_MAX or IFN_REDUC_MIN.
2964 2017-12-18 Jeff Law <law@redhat.com>
2966 * tree-ssa-dom.c (record_equivalences_from_phis): Do not
2967 record symbolic equivalences from backedges in the CFG.
2970 2017-11-19 Jeff Law <law@redhat.com>
2972 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
2973 of degenerates resulting from ignoring an edge.
2975 2017-12-18 Martin Sebor <msebor@redhat.com>
2978 PR tree-optimization/78450
2979 * tree-ssa-strlen.c (maybe_set_strlen_range): New function.
2980 (handle_builtin_strlen): Call it.
2982 2017-12-18 Segher Boessenkool <segher@kernel.crashing.org>
2984 PR rtl-optimization/83424
2985 * rtlanal.c (dead_or_set_regno_p): Handle CLOBBER just like SET.
2987 2017-12-18 Marek Polacek <polacek@redhat.com>
2990 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref):
2991 Check if TYPE is INTEGRAL_TYPE_P before accessing its min/max
2994 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
2996 * config/arc/arc.md (maddsidi4, maddsidi4_split): Update pattern.
2997 (umaddsidi4, umaddsidi_split): Likewise.
2999 2017-12-18 Claudiu Zissulescu <claziss@synopsys.com>
3001 * config/arc/arc.c (arc_legitimate_constant_p): Always check all
3004 2017-12-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3007 * config/s390/s390.c (s390_option_override): Avoid strncpy.
3009 2017-12-18 Richard Biener <rguenther@suse.de>
3011 PR tree-optimization/81877
3012 * tree-ssa-loop-im.c (ref_indep_loop_p): Remove safelen parameters.
3013 (outermost_indep_loop): Adjust.
3014 (ref_indep_loop_p_1): Likewise. Remove safelen handling again.
3015 (can_sm_ref_p): Adjust.
3017 2017-12-18 Richard Biener <rguenther@suse.de>
3020 * tree.c (array_at_struct_end_p): Return true if the underlying
3021 object has space for at least one element in excess of what
3022 the array domain specifies.
3024 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3026 * doc/extend.texi (x86 Function Attributes): Reformat nocf_check
3027 example to avoid overfull hbox.
3028 * doc/invoke.texi (Option Summary): Add missing @gol.
3029 (C++ Dialect Options): Reformat -Wnoexcept-type example to avoid
3032 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3033 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3035 * doc/invoke.texi (Option Summary): Add -mverbose-cost-dump
3036 to AArch64 and ARM lists, plus missing -mflip-thumb for ARM.
3037 (AArch64 Options): Document -mverbose-cost-dump.
3038 (ARM Options): Likewise, plus -mflip-thumb.
3040 2017-12-17 Martin Sebor <msebor@redhat.com>
3043 * gimple-ssa-warn-restrict.c
3044 (builtin_memref::offset_out_of_bounds): Correct the handling of
3047 2017-12-17 Sandra Loosemore <sandra@codesourcery.com>
3049 * doc/invoke.texi (ARC Options): Add missing -mlra entry.
3051 2017-12-17 John David Anglin <danglin@gcc.gnu.org>
3053 * config/pa/pa.c (pa_som_asm_init_sections): Fix comment.
3055 2017-12-17 Markus Trippelsdorf <markus@trippelsdorf.de>
3057 * x86-tune-costs.h (skylake_cost, core_cost): Decrease r64 multiply
3060 2017-12-16 Sandra Loosemore <sandra@codesourcery.com>
3062 * doc/invoke.texi: Fix some typos.
3064 2017-12-16 Martin Sebor <msebor@redhat.com>
3066 PR tree-optimization/78918
3067 * Makefile.in (OBJS): Add gimple-ssa-warn-restrict.o.
3068 * builtins.c (check_sizes): Rename...
3069 (check_access): ...to this. Rename function arguments for clarity.
3070 (check_memop_sizes): Adjust names.
3071 (expand_builtin_memchr, expand_builtin_memcpy): Same.
3072 (expand_builtin_memmove, expand_builtin_mempcpy): Same.
3073 (expand_builtin_strcat, expand_builtin_stpncpy): Same.
3074 (check_strncat_sizes, expand_builtin_strncat): Same.
3075 (expand_builtin_strncpy, expand_builtin_memset): Same.
3076 (expand_builtin_bzero, expand_builtin_memcmp): Same.
3077 (expand_builtin_memory_chk, maybe_emit_chk_warning): Same.
3078 (maybe_emit_sprintf_chk_warning): Same.
3079 (expand_builtin_strcpy): Adjust.
3080 (expand_builtin_stpcpy): Same.
3081 (expand_builtin_with_bounds): Detect out-of-bounds accesses
3082 in pointer-checking forms of memcpy, memmove, and mempcpy.
3083 (gcall_to_tree_minimal, max_object_size): Define new functions.
3084 * builtins.h (max_object_size): Declare.
3085 * calls.c (alloc_max_size): Call max_object_size instead of
3086 hardcoding ssizetype limit.
3087 (get_size_range): Handle new argument.
3088 * calls.h (get_size_range): Add a new argument.
3089 * cfgexpand.c (expand_call_stmt): Propagate no-warning bit.
3090 * doc/invoke.texi (-Wrestrict): Adjust, add example.
3091 * gimple-fold.c (gimple_fold_builtin_memory_op): Detect overlapping
3093 (gimple_fold_builtin_memory_chk): Same.
3094 (gimple_fold_builtin_stxcpy_chk): New function.
3095 * gimple-ssa-warn-restrict.c: New source.
3096 * gimple-ssa-warn-restrict.h: New header.
3097 * gimple.c (gimple_build_call_from_tree): Propagate location.
3098 * passes.def (pass_warn_restrict): Add new pass.
3099 * tree-pass.h (make_pass_warn_restrict): Declare.
3100 * tree-ssa-strlen.c (handle_builtin_strcpy): Detect overlapping
3102 (handle_builtin_strcat): Same.
3103 (strlen_optimize_stmt): Rename...
3104 (strlen_check_and_optimize_stmt): ...to this. Handle strncat,
3105 stpncpy, strncpy, and their checking forms.
3107 2017-12-16 Jan Hubicka <hubicka@ucw.cz>
3109 PR rtl-optimization/82849
3110 * modulo-sched.c (sms_schedule): Use get_estimated_loop_iterations_int
3111 and get_max_loop_iterations_int.
3113 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3114 Alan Hayward <alan.hayward@arm.com>
3115 David Sherwood <david.sherwood@arm.com>
3117 * machmode.h (mode_for_size, int_mode_for_size, float_mode_for_size)
3118 (smallest_mode_for_size, smallest_int_mode_for_size): Take the mode
3119 size as a poly_uint64.
3120 (mode_for_vector, mode_for_int_vector): Take the number of vector
3121 elements as a poly_uint64.
3122 * stor-layout.c (mode_for_size, smallest_mode_for_size): Take the mode
3123 size as a poly_uint64.
3124 (mode_for_vector, mode_for_int_vector): Take the number of vector
3125 elements as a poly_uint64.
3127 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3128 Alan Hayward <alan.hayward@arm.com>
3129 David Sherwood <david.sherwood@arm.com>
3131 * machmode.h (MACRO_MODE): New macro.
3132 * addresses.h (base_reg_class, ok_for_base_p_1): Use it.
3133 * targhooks.c (default_libcall_value, default_secondary_reload)
3134 (default_memory_move_cost, default_register_move_cost)
3135 (default_class_max_nregs): Likewise.
3137 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3138 Alan Hayward <alan.hayward@arm.com>
3139 David Sherwood <david.sherwood@arm.com>
3141 * doc/sourcebuild.texi: Document IN_TARGET_CODE.
3142 * genattrtab.c (write_header): Define IN_TARGET_CODE to 1 in the
3144 * genautomata.c (main): Likewise.
3145 * genconditions.c (write_header): Likewise.
3146 * genemit.c (main): Likewise.
3147 * genextract.c (print_header): Likewise.
3148 * genopinit.c (main): Likewise.
3149 * genoutput.c (output_prologue): Likewise.
3150 * genpeep.c (main): Likewise.
3151 * genpreds.c (write_insn_preds_c): Likewise.
3152 * genrecog.c (writer_header): Likewise.
3153 * config/aarch64/aarch64-builtins.c (IN_TARGET_CODE): Define.
3154 * config/aarch64/aarch64-c.c (IN_TARGET_CODE): Likewise.
3155 * config/aarch64/aarch64.c (IN_TARGET_CODE): Likewise.
3156 * config/aarch64/cortex-a57-fma-steering.c (IN_TARGET_CODE): Likewise.
3157 * config/aarch64/driver-aarch64.c (IN_TARGET_CODE): Likewise.
3158 * config/alpha/alpha.c (IN_TARGET_CODE): Likewise.
3159 * config/alpha/driver-alpha.c (IN_TARGET_CODE): Likewise.
3160 * config/arc/arc-c.c (IN_TARGET_CODE): Likewise.
3161 * config/arc/arc.c (IN_TARGET_CODE): Likewise.
3162 * config/arc/driver-arc.c (IN_TARGET_CODE): Likewise.
3163 * config/arm/aarch-common.c (IN_TARGET_CODE): Likewise.
3164 * config/arm/arm-builtins.c (IN_TARGET_CODE): Likewise.
3165 * config/arm/arm-c.c (IN_TARGET_CODE): Likewise.
3166 * config/arm/arm.c (IN_TARGET_CODE): Likewise.
3167 * config/arm/driver-arm.c (IN_TARGET_CODE): Likewise.
3168 * config/avr/avr-c.c (IN_TARGET_CODE): Likewise.
3169 * config/avr/avr-devices.c (IN_TARGET_CODE): Likewise.
3170 * config/avr/avr-log.c (IN_TARGET_CODE): Likewise.
3171 * config/avr/avr.c (IN_TARGET_CODE): Likewise.
3172 * config/avr/driver-avr.c (IN_TARGET_CODE): Likewise.
3173 * config/avr/gen-avr-mmcu-specs.c (IN_TARGET_CODE): Likewise.
3174 * config/bfin/bfin.c (IN_TARGET_CODE): Likewise.
3175 * config/c6x/c6x.c (IN_TARGET_CODE): Likewise.
3176 * config/cr16/cr16.c (IN_TARGET_CODE): Likewise.
3177 * config/cris/cris.c (IN_TARGET_CODE): Likewise.
3178 * config/darwin.c (IN_TARGET_CODE): Likewise.
3179 * config/epiphany/epiphany.c (IN_TARGET_CODE): Likewise.
3180 * config/epiphany/mode-switch-use.c (IN_TARGET_CODE): Likewise.
3181 * config/epiphany/resolve-sw-modes.c (IN_TARGET_CODE): Likewise.
3182 * config/fr30/fr30.c (IN_TARGET_CODE): Likewise.
3183 * config/frv/frv.c (IN_TARGET_CODE): Likewise.
3184 * config/ft32/ft32.c (IN_TARGET_CODE): Likewise.
3185 * config/h8300/h8300.c (IN_TARGET_CODE): Likewise.
3186 * config/i386/djgpp.c (IN_TARGET_CODE): Likewise.
3187 * config/i386/driver-i386.c (IN_TARGET_CODE): Likewise.
3188 * config/i386/driver-mingw32.c (IN_TARGET_CODE): Likewise.
3189 * config/i386/host-cygwin.c (IN_TARGET_CODE): Likewise.
3190 * config/i386/host-i386-darwin.c (IN_TARGET_CODE): Likewise.
3191 * config/i386/host-mingw32.c (IN_TARGET_CODE): Likewise.
3192 * config/i386/i386-c.c (IN_TARGET_CODE): Likewise.
3193 * config/i386/i386.c (IN_TARGET_CODE): Likewise.
3194 * config/i386/intelmic-mkoffload.c (IN_TARGET_CODE): Likewise.
3195 * config/i386/msformat-c.c (IN_TARGET_CODE): Likewise.
3196 * config/i386/winnt-cxx.c (IN_TARGET_CODE): Likewise.
3197 * config/i386/winnt-stubs.c (IN_TARGET_CODE): Likewise.
3198 * config/i386/winnt.c (IN_TARGET_CODE): Likewise.
3199 * config/i386/x86-tune-sched-atom.c (IN_TARGET_CODE): Likewise.
3200 * config/i386/x86-tune-sched-bd.c (IN_TARGET_CODE): Likewise.
3201 * config/i386/x86-tune-sched-core.c (IN_TARGET_CODE): Likewise.
3202 * config/i386/x86-tune-sched.c (IN_TARGET_CODE): Likewise.
3203 * config/ia64/ia64-c.c (IN_TARGET_CODE): Likewise.
3204 * config/ia64/ia64.c (IN_TARGET_CODE): Likewise.
3205 * config/iq2000/iq2000.c (IN_TARGET_CODE): Likewise.
3206 * config/lm32/lm32.c (IN_TARGET_CODE): Likewise.
3207 * config/m32c/m32c-pragma.c (IN_TARGET_CODE): Likewise.
3208 * config/m32c/m32c.c (IN_TARGET_CODE): Likewise.
3209 * config/m32r/m32r.c (IN_TARGET_CODE): Likewise.
3210 * config/m68k/m68k.c (IN_TARGET_CODE): Likewise.
3211 * config/mcore/mcore.c (IN_TARGET_CODE): Likewise.
3212 * config/microblaze/microblaze-c.c (IN_TARGET_CODE): Likewise.
3213 * config/microblaze/microblaze.c (IN_TARGET_CODE): Likewise.
3214 * config/mips/driver-native.c (IN_TARGET_CODE): Likewise.
3215 * config/mips/frame-header-opt.c (IN_TARGET_CODE): Likewise.
3216 * config/mips/mips.c (IN_TARGET_CODE): Likewise.
3217 * config/mmix/mmix.c (IN_TARGET_CODE): Likewise.
3218 * config/mn10300/mn10300.c (IN_TARGET_CODE): Likewise.
3219 * config/moxie/moxie.c (IN_TARGET_CODE): Likewise.
3220 * config/msp430/driver-msp430.c (IN_TARGET_CODE): Likewise.
3221 * config/msp430/msp430-c.c (IN_TARGET_CODE): Likewise.
3222 * config/msp430/msp430.c (IN_TARGET_CODE): Likewise.
3223 * config/nds32/nds32-cost.c (IN_TARGET_CODE): Likewise.
3224 * config/nds32/nds32-fp-as-gp.c (IN_TARGET_CODE): Likewise.
3225 * config/nds32/nds32-intrinsic.c (IN_TARGET_CODE): Likewise.
3226 * config/nds32/nds32-isr.c (IN_TARGET_CODE): Likewise.
3227 * config/nds32/nds32-md-auxiliary.c (IN_TARGET_CODE): Likewise.
3228 * config/nds32/nds32-memory-manipulation.c (IN_TARGET_CODE): Likewise.
3229 * config/nds32/nds32-pipelines-auxiliary.c (IN_TARGET_CODE): Likewise.
3230 * config/nds32/nds32-predicates.c (IN_TARGET_CODE): Likewise.
3231 * config/nds32/nds32.c (IN_TARGET_CODE): Likewise.
3232 * config/nios2/nios2.c (IN_TARGET_CODE): Likewise.
3233 * config/nvptx/mkoffload.c (IN_TARGET_CODE): Likewise.
3234 * config/nvptx/nvptx.c (IN_TARGET_CODE): Likewise.
3235 * config/pa/pa.c (IN_TARGET_CODE): Likewise.
3236 * config/pdp11/pdp11.c (IN_TARGET_CODE): Likewise.
3237 * config/powerpcspe/driver-powerpcspe.c (IN_TARGET_CODE): Likewise.
3238 * config/powerpcspe/host-darwin.c (IN_TARGET_CODE): Likewise.
3239 * config/powerpcspe/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3240 * config/powerpcspe/powerpcspe-c.c (IN_TARGET_CODE): Likewise.
3241 * config/powerpcspe/powerpcspe-linux.c (IN_TARGET_CODE): Likewise.
3242 * config/powerpcspe/powerpcspe.c (IN_TARGET_CODE): Likewise.
3243 * config/riscv/riscv-builtins.c (IN_TARGET_CODE): Likewise.
3244 * config/riscv/riscv-c.c (IN_TARGET_CODE): Likewise.
3245 * config/riscv/riscv.c (IN_TARGET_CODE): Likewise.
3246 * config/rl78/rl78-c.c (IN_TARGET_CODE): Likewise.
3247 * config/rl78/rl78.c (IN_TARGET_CODE): Likewise.
3248 * config/rs6000/driver-rs6000.c (IN_TARGET_CODE): Likewise.
3249 * config/rs6000/host-darwin.c (IN_TARGET_CODE): Likewise.
3250 * config/rs6000/host-ppc64-darwin.c (IN_TARGET_CODE): Likewise.
3251 * config/rs6000/rs6000-c.c (IN_TARGET_CODE): Likewise.
3252 * config/rs6000/rs6000-linux.c (IN_TARGET_CODE): Likewise.
3253 * config/rs6000/rs6000-p8swap.c (IN_TARGET_CODE): Likewise.
3254 * config/rs6000/rs6000-string.c (IN_TARGET_CODE): Likewise.
3255 * config/rs6000/rs6000.c (IN_TARGET_CODE): Likewise.
3256 * config/rx/rx.c (IN_TARGET_CODE): Likewise.
3257 * config/s390/driver-native.c (IN_TARGET_CODE): Likewise.
3258 * config/s390/s390-c.c (IN_TARGET_CODE): Likewise.
3259 * config/s390/s390.c (IN_TARGET_CODE): Likewise.
3260 * config/sh/sh-c.c (IN_TARGET_CODE): Likewise.
3261 * config/sh/sh-mem.cc (IN_TARGET_CODE): Likewise.
3262 * config/sh/sh.c (IN_TARGET_CODE): Likewise.
3263 * config/sh/sh_optimize_sett_clrt.cc (IN_TARGET_CODE): Likewise.
3264 * config/sh/sh_treg_combine.cc (IN_TARGET_CODE): Likewise.
3265 * config/sparc/driver-sparc.c (IN_TARGET_CODE): Likewise.
3266 * config/sparc/sparc-c.c (IN_TARGET_CODE): Likewise.
3267 * config/sparc/sparc.c (IN_TARGET_CODE): Likewise.
3268 * config/spu/spu-c.c (IN_TARGET_CODE): Likewise.
3269 * config/spu/spu.c (IN_TARGET_CODE): Likewise.
3270 * config/stormy16/stormy16.c (IN_TARGET_CODE): Likewise.
3271 * config/tilegx/mul-tables.c (IN_TARGET_CODE): Likewise.
3272 * config/tilegx/tilegx-c.c (IN_TARGET_CODE): Likewise.
3273 * config/tilegx/tilegx.c (IN_TARGET_CODE): Likewise.
3274 * config/tilepro/mul-tables.c (IN_TARGET_CODE): Likewise.
3275 * config/tilepro/tilepro-c.c (IN_TARGET_CODE): Likewise.
3276 * config/tilepro/tilepro.c (IN_TARGET_CODE): Likewise.
3277 * config/v850/v850-c.c (IN_TARGET_CODE): Likewise.
3278 * config/v850/v850.c (IN_TARGET_CODE): Likewise.
3279 * config/vax/vax.c (IN_TARGET_CODE): Likewise.
3280 * config/visium/visium.c (IN_TARGET_CODE): Likewise.
3281 * config/vms/vms-c.c (IN_TARGET_CODE): Likewise.
3282 * config/vms/vms-f.c (IN_TARGET_CODE): Likewise.
3283 * config/vms/vms.c (IN_TARGET_CODE): Likewise.
3284 * config/xtensa/xtensa.c (IN_TARGET_CODE): Likewise.
3286 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3287 Alan Hayward <alan.hayward@arm.com>
3288 David Sherwood <david.sherwood@arm.com>
3290 * tree-ssa-sccvn.c (vn_reference_lookup_3): Avoid repeated
3293 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3294 Alan Hayward <alan.hayward@arm.com>
3295 David Sherwood <david.sherwood@arm.com>
3297 * doc/generic.texi (VEC_SERIES_EXPR): Document.
3298 * doc/md.texi (vec_series@var{m}): Document.
3299 * tree.def (VEC_SERIES_EXPR): New tree code.
3300 * tree.h (build_vec_series): Declare.
3301 * tree.c (build_vec_series): New function.
3302 * cfgexpand.c (expand_debug_expr): Handle VEC_SERIES_EXPR.
3303 * tree-pretty-print.c (dump_generic_node): Likewise.
3304 * gimple-pretty-print.c (dump_binary_rhs): Likewise.
3305 * tree-inline.c (estimate_operator_cost): Likewise.
3306 * expr.c (expand_expr_real_2): Likewise.
3307 * optabs-tree.c (optab_for_tree_code): Likewise.
3308 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
3309 * fold-const.c (const_binop): Fold VEC_SERIES_EXPRs of constants.
3310 * expmed.c (make_tree): Handle VEC_SERIES.
3311 * optabs.def (vec_series_optab): New optab.
3312 * optabs.h (expand_vec_series_expr): Declare.
3313 * optabs.c (expand_vec_series_expr): New function.
3314 * tree-vect-generic.c (expand_vector_operations_1): Check that
3315 the operands also have vector type.
3317 2017-12-16 Richard Sandiford <richard.sandiford@linaro.org>
3318 Alan Hayward <alan.hawyard@arm.com>
3319 David Sherwood <david.sherwood@arm.com>
3321 * doc/generic.texi (VEC_DUPLICATE_EXPR): Document.
3322 (VEC_COND_EXPR): Add missing @tindex.
3323 * doc/md.texi (vec_duplicate@var{m}): Document.
3324 * tree.def (VEC_DUPLICATE_EXPR): New tree codes.
3325 * tree.c (build_vector_from_val): Add stubbed-out handling of
3326 variable-length vectors, using VEC_DUPLICATE_EXPR.
3327 (uniform_vector_p): Handle VEC_DUPLICATE_EXPR.
3328 * cfgexpand.c (expand_debug_expr): Likewise.
3329 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
3330 * tree-inline.c (estimate_operator_cost): Likewise.
3331 * tree-pretty-print.c (dump_generic_node): Likewise.
3332 * tree-vect-generic.c (ssa_uniform_vector_p): Likewise.
3333 * fold-const.c (const_unop): Fold VEC_DUPLICATE_EXPRs of a constant.
3334 (test_vec_duplicate_folding): New function.
3335 (fold_const_c_tests): Call it.
3336 * optabs.def (vec_duplicate_optab): New optab.
3337 * optabs-tree.c (optab_for_tree_code): Handle VEC_DUPLICATE_EXPR.
3338 * optabs.h (expand_vector_broadcast): Declare.
3339 * optabs.c (expand_vector_broadcast): Make non-static. Try using
3340 vec_duplicate_optab.
3341 * expr.c (store_constructor): Try using vec_duplicate_optab for
3343 (expand_expr_real_2): Handle VEC_DUPLICATE_EXPR.
3345 2017-12-15 Markus Trippelsdorf <markus@trippelsdorf.de>
3348 * config/i386/x86-tune-costs.h (skylake_cost, core_cost): Increase
3349 div/mod latencies a bit.
3351 2017-12-15 Jeff Law <law@redhat.com>
3353 PR tree-optimization/36550
3354 * tree-ssa-threadupdate.c (count_stmts_and_phis_in_block): New.
3355 (mark_threaded_blocks): Rewrite code to avoid block copying when
3356 optimizing for size. Don't pessimize blocks which will be
3357 copied, but all the statements will be dead.
3359 2017-12-15 Alexandre Oliva <aoliva@redhat.com>
3361 PR tree-optimization/81165
3362 * tree-ssa-threadupdate.c (uses_in_bb): New.
3363 (estimate_threading_killed_stmts): New.
3364 * tree-ssa-threadupdate.h (estimate_threading_killed_stmts): Prototype.
3365 * tree-ssa-threadedge.c
3366 (record_temporary_equivalences_from_stmts_at_dest): Expand limit
3369 2017-12-15 Jeff Law <law@redhat.com>
3371 PR tree-optimization/83410
3372 * tree-ssa-threadupdate.c (thread_block_1): Avoid certain jump
3373 threads when parallelizing loops.
3375 2017-12-15 Jakub Jelinek <jakub@redhat.com>
3377 * tree-core.h (struct attribute_spec): Swap affects_type_identity and
3379 * config/alpha/alpha.c (vms_attribute_table): Swap
3380 affects_type_identity and handler fields, adjust comments.
3381 * config/mips/mips.c (mips_attribute_table): Likewise.
3382 * config/visium/visium.c (visium_attribute_table): Likewise.
3383 * config/epiphany/epiphany.c (epiphany_attribute_table): Likewise.
3384 * config/microblaze/microblaze.c (microblaze_attribute_table):
3386 * config/spu/spu.c (spu_attribute_table): Likewise.
3387 * config/mcore/mcore.c (mcore_attribute_table): Likewise.
3388 * config/arc/arc.c (arc_attribute_table): Likewise.
3389 * config/m68k/m68k.c (m68k_attribute_table): Likewise.
3390 * config/v850/v850.c (v850_handle_interrupt_attribute,
3391 v850_handle_data_area_attribute): Formatting fixes.
3392 (v850_attribute_table): Swap affects_type_identity and handler
3393 fields, adjust comments.
3394 * config/m32r/m32r.c (m32r_attribute_table): Likewise.
3395 * config/arm/arm.c (arm_attribute_table): Likewise.
3396 * config/avr/avr.c (avr_attribute_table): Likewise.
3397 * config/s390/s390.c (s390_attribute_table): Likewise.
3398 * config/sh/sh.c (sh_attribute_table): Likewise.
3399 * config/i386/i386.c (ix86_handle_cconv_attribute,
3400 ix86_handle_callee_pop_aggregate_return): Formatting fixes.
3401 (ix86_attribute_table): Swap affects_type_identity and handler
3402 fields, adjust comments.
3403 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3404 * config/sparc/sparc.c (sparc_attribute_table): Likewise.
3405 * config/m32c/m32c.c (m32c_attribute_table): Likewise.
3406 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Likewise.
3407 * config/ia64/ia64.c (ia64_attribute_table): Likewise.
3408 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
3409 * config/rx/rx.c (rx_attribute_table): Likewise.
3410 * config/cr16/cr16.c (cr16_attribute_table): Likewise.
3411 * config/h8300/h8300.c (h8300_attribute_table): Likewise.
3412 * config/nvptx/nvptx.c (nvptx_attribute_table): Likewise.
3413 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Likewise.
3414 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
3415 * config/stormy16/stormy16.c (xstormy16_attribute_table): Likewise.
3416 * config/bfin/bfin.c (bfin_attribute_table): Likewise.
3417 * config/rs6000/rs6000.c (rs6000_attribute_table): Likewise.
3418 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
3419 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
3420 * doc/plugins.texi (user_attr): Likewise. Add NULL for
3422 * attribs.c (empty_attribute_table): Swap affects_type_identity and
3424 (register_scoped_attributes, decl_attributes): Formatting fixes.
3426 PR tree-optimization/83269
3427 * fold-const.c (fold_binary_loc): Perform (-A) - B -> (-B) - A
3428 subtraction in arg0's type if type is signed and arg0 is unsigned.
3432 * match.pd ((T)(P + A) - (T)P -> (T) A): Use @@0 instead of @0 and
3433 convert? on @0 instead of convert. Check type of @1, not @0.
3434 ((T)P - (T)(P + A) -> -(T) A): Use @@0 instead of @0 and
3435 convert? on @0 instead of convert. Check type of @1, not @0.
3436 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Use @@0 instead of @0,
3437 only optimize if either both @1 and @2 types are narrower
3438 precision, or both are wider or equal precision, and in the former
3439 case only if both have undefined overflow.
3441 2017-12-15 Richard Biener <rguenther@suse.de>
3444 * internal-fn.def (IFN_NOP): Add.
3445 * internal-fn.c (expand_NOP): Do nothing.
3446 * lto-streamer-in.c (input_function): Instead of removing
3447 sanitizer calls replace them with IFN_NOP calls.
3449 2017-12-15 Richard Sandiford <richard.sandiford@linaro.org>
3450 Alan Hayward <alan.hayward@arm.com>
3451 David Sherwood <david.sherwood@arm.com>
3453 * dse.c (store_info, read_info_type): Replace begin and end with
3455 (print_range): New function.
3456 (set_all_positions_unneeded, any_positions_needed_p)
3457 (check_mem_read_rtx, scan_stores, scan_reads, dse_step5): Update
3459 (record_store): Likewise. Optimize the case in which all positions
3461 (get_stored_val): Replace read_begin and read_end with read_offset
3463 (replace_read): Update call accordingly.
3465 2017-12-15 Bin Cheng <bin.cheng@arm.com>
3467 * gimple-loop-interchange.cc (STMT_COST_RATIO): New macro.
3468 (loop_cand::m_num_stmts, loop_cand::m_const_init_reduc): New members.
3469 (loop_cand::loop_cand): Initialize above members.
3470 (loop_cand::supported_operations): Delete.
3471 (loop_cand::can_interchange_p): Inline above function.
3472 (loop_cand::classify_simple_reduction): Record number of constant
3473 initialized simple reductions.
3474 (should_interchange_loops): New parameters. Check stmt cost of loops
3476 (tree_loop_interchange::interchange): Prepare stmt cost of outer loop.
3477 Update call to should_interchange_loops.
3478 (should_interchange_loop_nest): Update call to
3479 should_interchange_loops.
3481 2017-12-15 Eric Botcazou <ebotcazou@adacore.com>
3484 * ggc-page.c (HOST_BITS_PER_PTR): Do not define here...
3485 * hwint.h (HOST_BITS_PER_PTR): ...but here instead.
3486 * config/i386/xm-mingw32.h (HOST_BITS_PER_PTR): Delete.
3488 2017-12-15 Julia Koval <julia.koval@intel.com>
3490 * config/i386/i386-builtin.def (__builtin_ia32_vaesenclast_v16qi,
3491 __builtin_ia32_vaesenclast_v32qi, __builtin_ia32_vaesenclast_v64qi):
3493 * config/i386/sse.md (vaesenclast_<mode>): New pattern.
3494 * config/i386/vaesintrin.h (_mm256_aesenclast_epi128,
3495 _mm512_aesenclast_epi128, _mm_aesenclast_epi128): New intrinsics.
3497 2017-12-15 Julia Koval <julia.koval@intel.com>
3499 * config/i386/i386-builtin.def (__builtin_ia32_vaesenc_v16qi,
3500 __builtin_ia32_vaesenc_v32qi, __builtin_ia32_vaesenc_v64qi): New.
3501 * config/i386/sse.md (vaesenc_<mode>): New pattern.
3502 * config/i386/vaesintrin.h (_mm256_aesenc_epi128, _mm512_aesenc_epi128,
3503 _mm_aesenc_epi128): New intrinsics.
3505 2017-12-15 Julia Koval <julia.koval@intel.com>
3507 * config/i386/i386-builtin.def (__builtin_ia32_vaesdeclast_v16qi,
3508 __builtin_ia32_vaesdeclast_v32qi, __builtin_ia32_vaesdeclast_v64qi):
3510 * config/i386/sse.md (vaesdeclast_<mode>): New pattern.
3511 * config/i386/vaesintrin.h (_mm256_aesdeclast_epi128,
3512 _mm512_aesdeclast_epi128, _mm_aesdeclast_epi128): New intrinsics.
3514 2017-12-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3516 * gimple-ssa-strength-reduction.c (analyze_increments):
3517 Distinguish replacement costs for constant strides from those for
3520 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3522 * var-tracking.c (variable_tracking_main_1): Formatting fix.
3524 2017-12-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
3526 * doc/invoke.texi: Document -Wcast-function-type.
3527 * recog.h (stored_funcptr): Change signature.
3528 * tree-dump.c (dump_node): Avoid warning.
3529 * typed-splay-tree.h (typed_splay_tree): Avoid warning.
3531 2017-12-14 Qing Zhao <qing.zhao@oracle.com>
3534 * gimple-fold.c (get_range_strlen): Add the handling of non-member
3537 2017-12-14 David Malcolm <dmalcolm@redhat.com>
3539 PR tree-optimization/83312
3540 * domwalk.h (dom_walker::dom_walker): Fix typo in comment.
3541 * tree-cfg.c (find_taken_edge): Update to handle NULL_TREE for
3542 "val" param, and to cope with arbitrary basic blocks.
3543 (find_taken_edge_cond_expr): Add "cond_stmt" param and use it to
3544 handle NULL_TREE for "val", dropping "bb" param.
3545 (find_taken_edge_switch_expr): Make "switch_stmt" param const and
3546 drop "bb" param. Handle NULL_TREE for "val".
3547 (find_case_label_for_value): Make "switch_stmt" param const.
3548 * tree-vrp.c (class check_array_bounds_dom_walker): New subclass
3550 (vrp_prop::check_all_array_refs): Reimplement as...
3551 (check_array_bounds_dom_walker::before_dom_children): ...this new
3552 vfunc. Replace linear search through BB block list, excluding
3553 those with non-executable in-edges via dominator walk.
3555 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3557 * config/arm/arm.opt (mverbose-cost-dump): New option.
3558 * config/arm/arm.c (arm_rtx_costs): Use it.
3560 2017-12-14 Andreas Schwab <schwab@linux-m68k.org>
3563 * reload1.c (emit_input_reload_insns): Skip debug markers.
3565 2017-12-14 Alexandre Oliva <aoliva@redhat.com>
3567 * config/i386/i386.c (rest_of_insert_endbranch): Use call loc
3571 * config/arc/arc.c (hwloop_optimize): Skip debug insns.
3572 * config/sh/sh-protos.h (sh_find_set_of_reg): Adjust.
3573 * config/sh/sh.c: Skip debug insns besides notes.
3574 * config/sh/sh.md: Likewise.
3575 * config/sh/sh_treg_combine.cc: Likewise.
3576 * config/sh/sync.md: Likewise.
3578 2017-12-14 Tom de Vries <tom@codesourcery.com>
3580 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
3581 Add item for weak_undefined.
3583 2017-12-14 Richard Biener <rguenther@suse.de>
3585 PR tree-optimization/67842
3586 * sese.h (bb_in_region): Remove #if 0'ed code.
3588 2017-12-14 Richard Biener <rguenther@suse.de>
3590 PR tree-optimization/83326
3591 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
3592 may_be_zero parameter and handle it by not marking the first
3593 peeled copy as not exiting the loop.
3594 (try_peel_loop): Likewise.
3595 (canonicalize_loop_induction_variables): Use number_of_iterations_exit
3596 to handle the case of constant or zero iterations and perform
3597 loop header copying on-the-fly.
3599 2017-12-14 Richard Biener <rguenther@suse.de>
3601 PR tree-optimization/83418
3602 * vr-values.c (vr_values::extract_range_for_var_from_comparison_expr):
3603 Instead of asserting we don't get unfolded comparisons deal with them.
3605 2017-12-14 Jakub Jelinek <jakub@redhat.com>
3608 * var-tracking.c (vt_initialize): Ignore non-DEBUG_INSNs outside of
3609 basic blocks. Assert debug bind insns don't appear outside of bbs,
3610 don't reset them. Assert insns without BLOCK_FOR_INSN are outside of
3613 PR tree-optimization/83198
3614 * gimple-ssa-sprintf.c (format_floating): Set type solely based on
3615 dir.modifier, regardless of TREE_TYPE (arg). Assume non-REAL_CST
3616 value if arg is a REAL_CST with incompatible type.
3618 2017-12-14 Sudakshina Das <sudi.das@arm.com>
3619 Bin Cheng <bin.cheng@arm.com>
3622 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Move LTGT
3624 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Add
3627 2017-12-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3629 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75,
3630 cortex-a75.cortex-a55): Specify AARCH64_FL_F16 in the arch features.
3632 2017-12-14 Richard Sandiford <richard.sandiford@linaro.org>
3633 Alan Hayward <alan.hayward@arm.com>
3634 David Sherwood <david.sherwood@arm.com>
3636 * poly-int.h: New file.
3637 * poly-int-types.h: Likewise.
3638 * coretypes.h: Include them.
3639 (POLY_INT_CONVERSION): Define.
3640 * target.def (estimated_poly_value): New hook.
3641 * doc/tm.texi.in (TARGET_ESTIMATED_POLY_VALUE): New hook.
3642 * doc/tm.texi: Regenerate.
3643 * doc/poly-int.texi: New file.
3644 * doc/gccint.texi: Include it.
3645 * doc/rtl.texi: Describe restrictions on subreg modes.
3646 * Makefile.in (TEXI_GCCINT_FILES): Add poly-int.texi.
3647 * genmodes.c (NUM_POLY_INT_COEFFS): Provide a default definition.
3648 (emit_insn_modes_h): Emit a definition of NUM_POLY_INT_COEFFS.
3649 * targhooks.h (default_estimated_poly_value): Declare.
3650 * targhooks.c (default_estimated_poly_value): New function.
3651 * target.h (estimated_poly_value): Likewise.
3652 * wide-int.h (WI_UNARY_RESULT): Use wi::binary_traits.
3653 (wi::unary_traits): Delete.
3654 (wi::binary_traits::signed_shift_result_type): Define for
3655 offset_int << HOST_WIDE_INT, etc.
3656 (generic_wide_int::operator <<=): Define for all types and use
3657 wi::lshift instead of <<.
3658 (wi::hwi_with_prec): Add a default constructor.
3659 (wi::ints_for): New class.
3660 (operator <<): Define for all wide-int types.
3661 (operator /): New function.
3662 (operator %): Likewise.
3663 * selftest.h (ASSERT_KNOWN_EQ, ASSERT_KNOWN_EQ_AT, ASSERT_MAYBE_NE)
3664 (ASSERT_MAYBE_NE_AT): New macros.
3666 2017-12-13 Eric Botcazou <ebotcazou@adacore.com>
3667 Dominik Vogt <vogt@linux.vnet.ibm.com>
3670 * emit-rtl.c (init_emit): Remove ??? comment.
3671 * explow.c (get_dynamic_stack_size): Take known alignment of stack
3672 pointer + STACK_DYNAMIC_OFFSET into account in lieu of STACK_BOUNDARY.
3673 * config/sparc/sparc.h (INIT_EXPANDERS): In 32-bit mode, lower the
3674 alignment of 3 virtual registers to BITS_PER_WORD.
3676 * config/sparc/sparc.c (sparc_compute_frame_size): Simplify.
3678 2017-12-13 Peter Bergner <bergner@vnet.ibm.com>
3680 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_HTM_NO_SUSPEND): New define.
3681 * config/rs6000/rs6000.c (cpu_supports_info): Use it.
3683 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3686 * reload1.c (eliminate_regs_in_insn): Skip debug markers.
3688 2017-12-13 Jakub Jelinek <jakub@redhat.com>
3690 * tree-cfg.c (verify_gimple_in_cfg): Verify no non-label stmts
3691 with the exception of debug begin stmt markers appear before
3695 * final.c (rest_of_handle_final): Call variable_tracking_main only
3696 if !flag_var_tracking.
3698 2017-12-13 Alexandre Oliva <aoliva@redhat.com>
3699 Jakub Jelinek <jakub@redhat.com>
3703 * tree-cfgcleanup.c (remove_forwarder_block): Keep after
3704 labels debug stmts that can only appear after labels.
3706 2017-12-13 Alexander Monakov <amonakov@ispras.ru>
3708 PR rtl-optimization/82398
3709 * sel-sched.c (sel_rank_for_schedule): Fix check for zero
3710 EXPR_USEFULNESS in priority comparison.
3712 2017-12-13 Segher Boessenkool <segher@kernel.crashing.org>
3714 PR rtl-optimization/83393
3715 * combine.c (move_deaths): If reg_stat points to a too new insn in
3716 last_death, do not use it: find the proper insn instead.
3718 2017-12-12 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3721 * config/rs6000/rs6000-string.c (expand_block_compare,
3722 expand_strn_compare): Fix set_mem_size() calls.
3724 2017-12-12 Jeff Law <law@redhat.com>
3726 PR tree-optimization/83298
3727 PR tree-optimization/83362
3728 PR tree-optimization/83383
3729 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Make
3730 push_value_range a public interface. Add new argument to
3731 record_ranges_from_stmt.
3732 * gimple-ssa-evrp-analyze.c
3733 (evrp_range_analyzer::record_ranges_from_stmt): Add new argument.
3734 Update comments. Handle recording temporary equivalences.
3735 * tree-ssa-dom.c (dom_opt_opt_walker::before_dom_children): Add
3736 new argument to call to evrp_range_analyzer::record_ranges_from_stmt.
3737 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Likewise.
3738 * tree-ssa-threadedge.c: Include alloc-pool.h, vr-values.h and
3739 gimple-ssa-evrp-analyze.h.
3740 (record_temporary_equivalences_from_phis): Add new argument. When
3741 the PHI arg is an SSA_NAME, set the result's range to the range
3743 (record_temporary_equivalences_from_stmts_at_dest): Record ranges
3744 from statements too.
3745 (thread_through_normal_block): Accept new argument, evrp_range_analyzer.
3746 Pass it down to children as needed.
3747 (thread_outgoing_edges): Likewise.
3748 (thread_across_edge): Likewise. Push/pop range state as needed.
3749 * tree-ssa-threadedge.h (thread_outgoing_edges): Update prototype.
3751 2017-12-12 Julia Koval <julia.koval@intel.com>
3753 * config/i386/i386.c (PTA_SKYLAKE_AVX512): Add PTA_CLWB.
3754 (PTA_CANNONLAKE): Remove PTA_CLWB.
3756 2017-12-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3759 * config/rs6000/vector.md (vcondv2dfv2di): New define_expand.
3760 (vcondv2div2df): Likewise.
3761 (vconduv2dfv2di): Likewise.
3763 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3765 * builtins.def (BUILT_IN_NEXTAFTER, BUILT_IN_NEXTAFTERF,
3766 BUILT_IN_NEXTAFTERL, BUILT_IN_NEXTTOWARD, BUILT_IN_NEXTTOWARDF,
3767 BUILT_IN_NEXTTOWARDL): Use ATTR_MATHFN_ERRNO instead of
3768 ATTR_MATHFN_FPROUNDING_ERRNO.
3770 2017-12-12 Richard Biener <rguenther@suse.de>
3772 PR tree-optimization/83385
3773 * graphite-scop-detection.c (get_order, order): Remove.
3774 (bb_to_rpo): New global.
3776 (build_scops): Sort pbbs in RPO order.
3778 2017-12-12 James Greenhalgh <james.greenhalgh@arm.com>
3780 * combine.c (simplify_set): Do not transform subregs to zero_extends
3781 if the destination mode is a vector mode.
3783 2017-12-12 Jakub Jelinek <jakub@redhat.com>
3785 PR tree-optimization/83359
3786 * tree-cfg.h (fold_loop_internal_call): Declare.
3787 * tree-vectorizer.c (fold_loop_internal_call): Moved to ...
3788 * tree-cfg.c (fold_loop_internal_call): ... here. No longer static.
3789 (find_loop_dist_alias): New function.
3790 (move_sese_region_to_fn): If any dloop->orig_loop_num value is
3791 updated, also adjust any corresponding LOOP_DIST_ALIAS internal
3794 PR tree-optimization/80631
3795 * tree-vect-loop.c (get_initial_def_for_reduction): Fix comment typo.
3796 (vect_create_epilog_for_reduction): Add INDUC_VAL and INDUC_CODE
3797 arguments, for INTEGER_INDUC_COND_REDUCTION use INDUC_VAL instead of
3798 hardcoding zero as the value if COND_EXPR is never true. For
3799 INTEGER_INDUC_COND_REDUCTION don't emit the final COND_EXPR if
3800 INDUC_VAL is equal to INITIAL_DEF, and use INDUC_CODE instead of
3801 hardcoding MAX_EXPR as the reduction operation.
3802 (is_nonwrapping_integer_induction): Allow negative step.
3803 (vectorizable_reduction): Compute INDUC_VAL and INDUC_CODE for
3804 vect_create_epilog_for_reduction, if no value is suitable, don't
3805 use INTEGER_INDUC_COND_REDUCTION for now. Formatting fixes.
3807 2017-12-12 Richard Biener <rguenther@suse.de>
3809 PR tree-optimization/81889
3810 * tree-ssa-loop-niter.c (infer_loop_bounds_from_signedness): Use
3811 range info from the non-wrapping IV instead of just the range
3814 2017-12-12 Julia Koval <julia.koval@intel.com>
3816 * config.gcc: Add vaesintrin.h.
3817 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
3819 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
3820 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
3822 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
3823 * config/i386/immintrin.h: Include vaesintrin.h.
3824 * config/i386/sse.md (vaesdec_<mode>): New pattern.
3825 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
3826 _mm_aesdec_epi128): New intrinsics.
3828 2017-12-12 Julia Koval <julia.koval@intel.com>
3830 * common/config/i386/i386-common.c (OPTION_MASK_ISA_VAES_SET,
3831 OPTION_MASK_ISA_VAES_UNSET): New.
3832 (ix86_handle_option): Handle -mvaes.
3833 * config/i386/cpuid.h: Define bit_VAES.
3834 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mvaes.
3835 * config/i386/i386-c.c (__VAES__): New.
3836 * config/i386/i386.c (ix86_target_string): Add -mvaes.
3837 (ix86_valid_target_attribute_inner_p): Ditto.
3838 * config/i386/i386.h (TARGET_VAES, TARGET_VAES_P): New.
3839 * config/i386/i386.opt: Add -mvaes.
3840 * doc/invoke.texi: Ditto.
3842 2017-12-12 Alexandre Oliva <aoliva@redhat.com>
3844 * debug.h (gcc_debug_hooks): Add inline_entry.
3845 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3846 * debug.c (do_nothing_debug_hooks): Likewise.
3847 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3848 * dwarf2out.c (dwarf2_debug_hooks): Likewise.
3849 (dwarf2_lineno_debug_hooks): Likewise.
3851 * common.opt (gstatement-frontiers): New, setting
3852 debug_nonbind_markers_p.
3853 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): Activate.
3854 * toplev.c (process_options): Autodetect value for debug statement
3856 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): Activate.
3857 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers): New.
3859 * cfgexpand.c (expand_gimple_basic_block): Handle begin stmt
3860 markers. Integrate source bind into debug stmt expand loop.
3861 (pass_expand::execute): Check debug marker limit. Avoid deep
3862 TER and expand debug locations for debug bind insns only.
3863 * cse.c (insn_live_p): Keep nonbind markers and debug bindings
3865 * df-scan.c (df_insn_delete): Accept out-of-block debug insn.
3866 * final.c (reemit_insn_block_notes): Take current block from
3867 nonbind markers. Declare note where it's first set.
3868 (final_scan_insn): Handle begin stmt notes. Emit is_stmt according to
3869 begin stmt markers if enabled.
3870 (notice_source_line): Handle nonbind markers. Fail if their
3871 location is unknown or that of builtins.
3872 (rest_of_handle_final): Convert begin stmt markers to notes if
3873 var-tracking didn't run.
3874 (rest_of_clean_state): Skip begin stmt markers.
3875 * gimple-pretty-print.c (dump_gimple_debug): Handle begin stmt
3877 * function.c (allocate_struct_function): Set begin_stmt_markers.
3878 * function.h (struct function): Add debug_marker_count counter
3879 and debug_nonbind_markers flag.
3880 * gimple-iterator.c (gsi_remove): Adjust debug_marker_count.
3881 * gimple-low.c (lower_function_body): Adjust
3882 debug_nonbind_markers.
3883 (lower_stmt): Drop or skip gimple debug stmts.
3884 (lower_try_catch): Skip debug stmts.
3885 * gimple.c (gimple_build_debug_begin_stmt): New.
3886 (gimple_copy): Increment debug_marker_count if copying one.
3887 * gimple.h (gimple_build_debug_begin_stmt): Declare.
3888 * gimplify.c (rexpr_location): New.
3889 (rexpr_has_location): New.
3890 (warn_switch_unreachable_r): Handle gimple debug stmts.
3891 (shortcut_cond_r): Call expr_location.
3893 (find_goto_label): New.
3894 (shortcut_cond_expr): Call expr_has_location, expr_location, and
3896 (gimplify_cond_expr): Call find_goto_label, expr_has_location, and
3898 (gimplify_expr): Handle begin stmt markers. Reject debug expr decls.
3899 * langhooks-def.h (LANG_HOOKS_EMITS_BEGIN_STMT): New. Add to...
3900 (LANG_HOOKS_INITIALIZER): ... this.
3901 * langhooks.h (struct lang_hooks): Add emits_begin_stmt.
3902 * lra-contraints.c (inherit_reload_reg): Tolerate between-blocks
3904 (update_ebb_live_info): Skip debug insn markers.
3905 * lra.c (debug_insn_static_data): Rename to...
3906 (debug_bind_static_data): ... this.
3907 (debug_marker_static_data): New.
3908 (lra_set_insn_recog_data): Select one of the above depending
3910 (lra_update_isn_regno_info): Don't assume debug insns have
3912 (push_insns): Skip debug insns.
3913 * lto-streamer-in.c (input_function): Drop debug stmts
3914 depending on active options. Adjust debug_nonbind_markers.
3915 * params.def (PARAM_MAX_DEBUG_MARKER_COUNT): New.
3916 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
3917 begin stmt marker notes.
3918 (print_insn): Likewise.
3919 * recog.c (extract_insn): Recognize rtl for debug markers.
3920 * rtl.def (DEBUG_MARKER): New.
3921 * tree-inline.c: Include params.h.
3922 (remap_gimple_stmt): Handle nonbind markers.
3923 (maybe_move_debug_stmts_to_successors): Likewise.
3924 (copy_debug_stmt): Likewise.
3925 * tree-iterator.c (append_to_statement_list_1): Append begin stmt
3926 markers regardless of no side effects.
3927 (tsi_link_before): Don't update container's side effects when adding
3928 a begin stmt marker.
3929 (tsi_link_after): Likewise.
3930 (expr_first): Skip begin stmt markers.
3931 (expr_last): Likewise.
3932 * tree-pretty-print (dump_generic_node): Handle begin stmt markers.
3933 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
3934 Disregard nonbind markers.
3935 * tree.c (make_node_stat): Don't set side effects for begin stmt
3937 (build1_stat): Likewise.
3938 * tree.def (DEBUG_BEGIN_STMT): New.
3939 * tree.h (GOTO_DESTINATION): Require a GOTO_EXPR.
3940 * var-tracking.c (delete_debug_insns): Renamed to...
3941 (delete_vta_debug_insns): ... this.
3942 (reemit_marker_as_note): New.
3943 (vt_initialize): Reemit markers.
3944 (delete_vta_debug_insns): Likewise.
3945 (vt_debug_insns_local): Reemit or delete markers.
3946 (variable_tracking_main_1): Likewise.
3947 * doc/generic.texi (DEBUG_BEGIN_STMT): Document.
3948 * doc/gimple.texi (gimple_debug_begin_stmt_p): New.
3949 (gimple_debug_nonbind_marker_p): New.
3950 (gimple_build_debug_bind): Adjust.
3951 (gimple_build_debug_begin_stmt): New.
3952 * doc/invoke.texi (max-debug-marker-count): New param.
3953 * doc/rtl.texi (debug_implicit_ptr, entry_value): New.
3954 (debug_parameter_ref, debug_marker): New.
3955 (NOTE_INSN_BEGIN_STMT): New.
3956 (DEBUG_INSN): Describe begin stmt markers.
3958 * cfgbuild.c (find_bb_boundaries): Don't purge dead edges if,
3959 without debug insns, we wouldn't, but clean up debug insns
3960 after a control flow insn nevertheless.
3962 * cfgcleanup.c (delete_unreachable_blocks): Use alternate
3963 block removal order if MAY_HAVE_DEBUG_BIND_INSNS.
3964 * cfgexpand.c (label_rtx_for_bb): Skip debug insns.
3965 * cfgrtl.c (try_redirect_by_replacing_jump): Skip debug insns.
3966 (rtl_tidy_fallthru_edge): Likewise.
3967 (rtl_verify_fallthru): Likewise.
3968 (rtl_verify_bb_layout): Likewise.
3969 (skip_insns_after_block): Likewise.
3970 (duplicate_insn_chain): Use DEBUG_BIND_INSN_P.
3971 * dwarf2out.c: Include print-rtl.h.
3972 (dwarf2out_next_real_insn): New.
3973 (dwarf2out_var_location): Call it. Disregard begin stmt markers.
3974 Dump debug binds in asm comments.
3975 * gimple-iterator.c (gimple_find_edge_insert_loc): Skip debug stmts.
3976 * gimple-iterator.h (gsi_start_bb_nondebug): Remove; adjust
3977 callers to use gsi_start_nondebug_bb instead.
3978 (gsi_after_labels): Skip gimple debug stmts.
3979 (gsi_start_nondebug): New.
3980 * gimple-loop-interchange.c (find_deps_in_bb_for_stmt): Adjust.
3981 (proper_loop_form_for_interchange): Adjust.
3982 * gimple-low.c (gimple_seq_may_fallthru): Take last nondebug stmt.
3983 * gimple.h (gimple_seq_last_nondebug_stmt): New.
3984 * gimplify.c (last_stmt_in_scope): Skip debug stmts.
3985 (collect_fallthrough_labels): Likewise.
3986 (should_warn_for_implicit_fallthrough): Likewise.
3987 (warn_implicit_fallthrough_r): Likewise.
3988 (expand_FALLTHROUGH_r): Likewise.
3989 * graphite-isl-ast-to-gimple.c (gsi_insert_earliest): Adjust.
3990 (graphite_copy_stmts_from_block): Skip nonbind markers.
3991 * haifa-sched.c (sched_extend_bb): Skip debug insns.
3992 * ipa-icf-gimple.c (func_checker::compare_bb): Adjust.
3993 * jump.c (clean_barriers): Skip debug insns.
3994 * omp-expand.c (expand_parallel_call): Skip debug insns.
3995 (expand_task_call): Likewise.
3996 (remove_exit_barrier): Likewise.
3997 (expand_omp_taskreg): Likewise.
3998 (expand_omp_for_init_counts): Likewise.
3999 (expand_omp_for_generic): Likewise.
4000 (expand_omp_for_static_nochunk): Likewise.
4001 (expand_omp_for_static_chunk): Likewise.
4002 (expand_omp_simd): Likewise.
4003 (expand_omp_taskloop_for_outer): Likewise.
4004 (expand_omp_taskloop_for_inner): Likewise.
4005 (expand_oacc_for): Likewise.
4006 (expand_omp_sections): Likewise.
4007 (expand_omp_single): Likewise.
4008 (expand_omp_synch): Likewise.
4009 (expand_omp_atomic_load): Likewise.
4010 (expand_omp_atomic_store): Likewise.
4011 (expand_omp_atomic_fetch_op): Likewise.
4012 (expand_omp_atomic_pipeline): Likewise.
4013 (expand_omp_atomic_mutex): Likewise.
4014 (expand_omp_target): Likewise.
4015 (grid_expand_omp_for_loop): Likewise.
4016 (grid_expand_target_grid_body): Likewise.
4017 (build_omp_regions_1): Likewise.
4018 * omp-low.c (check_combined_parallel): Skip debug stmts.
4019 * postreload.c (fixup_debug_insns): Skip nonbind debug insns.
4020 * regcprop.c (find_oldest_value_reg): Ensure REGNO is not a pseudo.
4021 * sese.c (sese_trivially_empty_bb_p): Call is_gimple_debug in
4023 * tree-cfg.c (make_blobs_1): Skip debug stmts.
4024 (make_edges): Likewise.
4025 (cleanup_dead_labels): Likewise.
4026 (gimple_can_merge_blocks_p): Likewise.
4027 (stmt_starts_bb_p): Likewise.
4028 (gimple_block_label): Likewise.
4029 (gimple_redirect_edge_and_branch): Likewise.
4030 * tree-cfgcleanup.c (remove_forwarder_block): Rearrange skipping
4032 (execute_cleanup_cfg_post_optimizing): Dump enumerated decls with
4034 * tree-pretty-print (print_declaration): Omit initializer in slim
4036 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Mark begin stmt
4038 (eliminate_unnecessary_stmts): Stabilize block removal order.
4039 * tree-ssa-tail-merge.c (find_duplicate): Skip debug stmts.
4040 * var-tracking.c (get_first_insn): New.
4041 (vt_emit_notes): Call it.
4042 (vt_initialize): Walk any insns before the first BB.
4043 (delete_debug_insns): Likewise.
4045 * gimple.h (enum gimple_debug_subcode): Add
4046 GIMPLE_DEBUG_BEGIN_STMT.
4047 (gimple_debug_begin_stmt_p): New.
4048 (gimple_debug_nonbind_marker_p): New.
4049 * tree.h (MAY_HAVE_DEBUG_MARKER_STMTS): New.
4050 (MAY_HAVE_DEBUG_BIND_STMTS): Renamed from....
4051 (MAY_HAVE_DEBUG_STMTS): ... this. Check both.
4052 * insn-notes.def (BEGIN_STMT): New.
4053 * rtl.h (MAY_HAVE_DEBUG_MARKER_INSNS): New.
4054 (MAY_HAVE_DEBUG_BIND_INSNS): Renamed from....
4055 (MAY_HAVE_DEBUG_INSNS): ... this. Check both.
4056 (NOTE_MARKER_LOCATION, NOTE_MARKER_P): New.
4057 (DEBUG_BIND_INSN_P, DEBUG_MARKER_INSN_P): New.
4058 (INSN_DEBUG_MARKER_KIND): New.
4059 (GEN_RTX_DEBUG_MARKER_BEGIN_STMT_PAT): New.
4060 (INSN_VAR_LOCATION): Check for VAR_LOCATION.
4061 (INSN_VAR_LOCATION_PTR): New.
4062 * cfgexpand.c (expand_debug_locations): Handle debug bind insns
4064 (expand_gimple_basic_block): Likewise. Emit debug temps for TER
4065 deps only if debug bind insns are enabled.
4066 (pass_expand::execute): Avoid deep TER and expand
4067 debug locations for debug bind insns only.
4068 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Narrow
4069 debug stmts special handling down to debug bind stmts.
4070 * combine.c (try_combine): Narrow debug insns special handling
4071 down to debug bind insns.
4072 * cse.c (delete_trivially_dead_insns): Handle debug bindings.
4073 Narrow debug insns preexisting special handling down to debug
4075 * dce.c (rest_of_handle_ud_dce): Narrow debug insns special
4076 handling down to debug bind insns.
4077 * function.c (instantiate_virtual_regs): Skip debug markers,
4078 adjust handling of debug binds.
4079 * gimple-ssa-backprop.c (backprop::prepare_change): Try debug
4080 temp insertion iff MAY_HAVE_DEBUG_BIND_STMTS.
4081 * haifa-sched.c (schedule_insn): Narrow special handling of debug
4082 insns to debug bind insns.
4083 * ipa-param-manipulation.c (ipa_modify_call_arguments): Narrow
4084 special handling of debug stmts to debug bind stmts.
4085 * ipa-split.c (split_function): Likewise.
4086 * ira.c (combine_and_move_insns): Adjust debug bind insns only.
4087 * loop-unroll.c (apply_opt_in_copies): Adjust tests on bind
4089 * reg-stack.c (convert_regs_1): Use DEBUG_BIND_INSN_P.
4090 * regrename.c (build_def_use): Likewise.
4091 * regcprop.c (copyprop_hardreg_forward_1): Likewise.
4092 (pass_cprop_hardreg): Narrow special casing of debug insns to
4094 * regstat.c (regstat_init_n_sets_and_refs): Likewise.
4095 * reload1.c (reload): Likewise.
4096 * sese.c (sese_insert_phis_for_liveouts): Narrow special
4097 casing of debug stmts to debug bind stmts.
4098 * shrink-wrap.c (move_insn_for_shrink_wrap): Likewise.
4099 * ssa-iterators.h (num_imm_uses): Likewise.
4100 * tree-cfg.c (gimple_merge_blocks): Narrow special casing of
4101 debug stmts to debug bind stmts.
4102 * tree-inline.c (tree_function_versioning): Narrow special casing
4103 of debug stmts to debug bind stmts.
4104 * tree-loop-distribution.c (generate_loops_for_partition):
4105 Narrow special casing of debug stmts to debug bind stmts.
4106 * tree-sra.c (analyze_access_subtree): Narrow special casing
4107 of debug stmts to debug bind stmts.
4108 * tree-ssa-dce.c (remove_dead_stmt): Narrow special casing of debug
4109 stmts to debug bind stmts.
4110 * tree-ssa-loop-ivopt.c (remove_unused_ivs): Narrow special
4111 casing of debug stmts to debug bind stmts.
4112 * tree-ssa-reassoc.c (reassoc_remove_stmt): Likewise.
4113 * tree-ssa-tail-merge.c (tail_merge_optimize): Narrow special
4114 casing of debug stmts to debug bind stmts.
4115 * tree-ssa-threadedge.c (propagate_threaded_block_debug_info):
4117 * tree-ssa.c (flush_pending_stmts): Narrow special casing of
4118 debug stmts to debug bind stmts.
4119 (gimple_replace_ssa_lhs): Likewise.
4120 (insert_debug_temp_for_var_def): Likewise.
4121 (insert_debug_temps_for_defs): Likewise.
4122 (reset_debug_uses): Likewise.
4123 * tree-ssanames.c (release_ssa_name_fn): Likewise.
4124 * tree-vect-loop-manip.c (adjust_debug_stmts_now): Likewise.
4125 (adjust_debug_stmts): Likewise.
4126 (adjust_phi_and_debug_stmts): Likewise.
4127 (vect_do_peeling): Likewise.
4128 * tree-vect-loop.c (vect_transform_loop): Likewise.
4129 * valtrack.c (propagate_for_debug): Use BIND_DEBUG_INSN_P.
4130 * var-tracking.c (adjust_mems): Narrow special casing of debug
4131 insns to debug bind insns.
4132 (dv_onepart_p, dataflow_set_clar_at_call, use_type): Likewise.
4133 (compute_bb_dataflow, vt_find_locations): Likewise.
4134 (vt_expand_loc, emit_notes_for_changes): Likewise.
4135 (vt_init_cfa_base): Likewise.
4136 (vt_emit_notes): Likewise.
4137 (vt_initialize): Likewise.
4138 (vt_finalize): Likewise.
4140 * emit-rtl.c (next_nondebug_insn, prev_nondebug_insn): Reorder.
4141 (next_nonnote_nondebug_insn, prev_nonnote_nondebug_insn): Reorder.
4142 (next_nonnote_nondebug_insn_bb): New.
4143 (prev_nonnote_nondebug_insn_bb): New.
4144 (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove.
4145 * rtl.h (prev_nonnote_insn_bb, next_nonnote_insn_bb): Remove decls.
4146 (prev_nonnote_nondebug_insn_bb): Declare.
4147 (next_nonnote_nondebug_insn_bb): Declare.
4148 * cfgbuild.c (find_bb_boundaries): Adjust to skip debug insns.
4149 * cfgrtl.c (get_last_bb_insn): Likewise.
4150 * lra.c (push_insns): Likewise.
4152 2017-12-11 David Malcolm <dmalcolm@redhat.com>
4155 * selftest-run-tests.c (selftest::run_tests): Move start/finish code
4157 * selftest.c (selftest::test_runner::test_runner): New ctor.
4158 (selftest::test_runner::~test_runner): New dtor.
4159 * selftest.h (class selftest::test_runner): New class.
4161 2017-12-11 Carl Love <cel@us.ibm.com>
4163 * config/rs6000/altivec.h (vec_extract_fp32_from_shorth,
4164 vec_extract_fp32_from_shortl]): Add #defines.
4165 * config/rs6000/rs6000-builtin.def (VSLDOI_2DI): Add macro expansion.
4166 * config/rs6000/rs6000-c.c (ALTIVEC_BUILTIN_VEC_UNPACKH,
4167 ALTIVEC_BUILTIN_VEC_UNPACKL, ALTIVEC_BUILTIN_VEC_AND,
4168 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SRL,
4169 ALTIVEC_BUILTIN_VEC_SRO, ALTIVEC_BUILTIN_VEC_SLD,
4170 ALTIVEC_BUILTIN_VEC_SLL): Add expansions.
4171 * doc/extend.texi: Add documentation for the added builtins.
4173 2017-12-11 Alexandre Oliva <aoliva@redhat.com>
4175 PR rtl-optimization/80693
4176 PR rtl-optimization/81019
4177 PR rtl-optimization/81020
4178 * combine.c (distribute_notes): Reset any REG_UNUSED REGs that
4179 are not mentioned in i3. Place the REG_UNUSED note on i2,
4180 possibly modified to REG_DEAD, if it did not originate in i3.
4182 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4184 * recog.c (store_data_bypass_p_1): New function.
4185 (store_data_bypass_p): Handle USE in a PARALLEL like CLOBBER. Use
4186 store_data_bypass_p_1 to avoid code duplication. Formatting fixes.
4188 2017-12-11 Segher Boessenkool <segher@kernel.crashing.org>
4190 PR rtl-optimization/83361
4191 * ifcvt.c (if_convert): Call fixup_partitions.
4193 2017-12-11 Will Schmidt <will_schmidt@vnet.ibm.com>
4195 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
4196 early folding of splat_u{8,16,32}.
4198 2017-12-11 Jakub Jelinek <jakub@redhat.com>
4200 * config/aarch64/aarch64.c (aarch64_print_operand): Don't start
4201 output_operand_lossage first argument with capital letter.
4202 (aarch64_override_options): Don't start error and sorry first argument
4203 with capital letter.
4205 2017-12-11 Andi Kleen <ak@linux.intel.com>
4207 PR gcov-profile/83355
4208 * auto-profile.c (string_table::get_index_by_decl): Don't
4209 recurse when abstract origin points to itself.
4211 2017-12-11 Bin Cheng <bin.cheng@arm.com>
4213 PR tree-optimization/83320
4214 * gimple-loop-interchange.cc (free_data_refs_with_aux): Use delete.
4215 (prune_datarefs_not_in_loop): Ditto.
4217 2017-12-10 Gerald Pfeifer <gerald@pfeifer.com>
4219 * doc/install.texi (Specific): Tweak link to mkssoftware.com.
4221 2017-12-10 Jakub Jelinek <jakub@redhat.com>
4223 PR tree-optimization/83337
4224 * gimple-loop-interchange.cc (compute_access_stride): Handle
4225 bitfield DRs properly.
4227 2017-12-09 Jakub Jelinek <jakub@redhat.com>
4229 PR tree-optimization/83338
4230 * tree-vect-stmts.c (vectorizable_operation): Handle POINTER_DIFF_EXPR
4231 vectorization as MINUS_EXPR with a subsequent VIEW_CONVERT_EXPR from
4232 vector of unsigned integers to vector of signed integers.
4234 2017-12-08 Vladimir Makarov <vmakarov@redhat.com>
4236 PR rtl-optimization/83317
4237 * lra-constraints.c (process_address_1): Add insn code check.
4239 2017-12-08 Michael Matz <matz@suse.de>
4241 Fix PR tree-optimization/83323
4242 * gimple-loop-jam (unroll_jam_possible_p): Correct test for
4243 head-controlled loops and loop BBs.
4244 * common.opt (funroll-and-jam): Remove, instead ...
4245 (floop-unroll-and-jam): ... reuse this option.
4246 * opts.c (default_options_table): Use OPT_floop_unroll_and_jam.
4247 * doc/invoke.texi (-funroll-and-jam): Move docu to ...
4248 (-floop-unroll-and-jam): ... this option.
4250 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4252 * ipa-polymorphic-call.c (noncall_stmt_may_be_vtbl_ptr_store): Fix
4253 a comment typo, get_base_ref_and_offset -> get_ref_base_and_extent.
4254 * ipa-prop.c (stmt_may_be_vtbl_ptr_store): Likewise.
4256 2017-12-08 Richard Biener <rguenther@suse.de>
4259 * tree-ssa-uninit.c (warn_uninitialized_vars): Properly
4260 handle accesses outside of zero-sized vars.
4262 2017-12-08 Martin Jambor <mjambor@suse.cz>
4264 PR tree-optimization/83141
4265 * tree-sra.c (contains_vce_or_bfcref_p): Move up in the file, also
4266 test for MEM_REFs implicitely changing types with padding. Remove
4268 (build_accesses_from_assign): Added contains_vce_or_bfcref_p checks.
4270 2017-12-08 Jakub Jelinek <jakub@redhat.com>
4272 * config/arc/arc.c (arc_attribute_table): Add exclusions to
4274 * config/avr/avr.c (avr_attribute_table): Likewise.
4275 * config/msp430/msp430.c (msp430_attribute_table): Likewise.
4276 * config/rl78/rl78.c (rl78_attribute_table): Likewise.
4277 * config/nds32/nds32.c (nds32_attribute_table): Likewise.
4278 * config/darwin.h (SUBTARGET_ATTRIBUTE_TABLE): Initialize new member
4279 of struct attribute_spec.
4280 * config/i386/cygming.h (SUBTARGET_ATTRIBUTE_TABLE): Likewise.
4282 2017-12-08 Ulrich Weigand <uweigand@de.ibm.com>
4285 * config/spu/spu.c (pad_bb): Only check INSN_CODE when INSN_P is true.
4287 2017-12-08 Jan Hubicka <hubicka@ucw.cz>
4290 * profile-count.c (profile_count::from_gcov_type): Move from
4291 profile-count.h; handle overflow.
4292 * profile-count.h (profile_count::from_gcov_type): Move offline.
4294 2017-12-08 Segher Boessenkool <segher@kernel.crashing.org>
4296 PR rtl-optimization/83304
4297 * combine.c (move_deaths): If we do not know where a register died,
4300 2017-12-08 Richard Biener <rguenther@suse.de>
4302 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
4303 Provide -fopt-info-loop feedback when we interchange in a nest.
4305 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4307 * config/arm/driver-arm.c (arm_cpu_table): Use fp-capable product names
4308 for armv6 ARM CPU IDs.
4310 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4312 * common/config/arm/arm-common.c: Include <algorithm>.
4313 (INCLUDE_VECTOR): Define.
4314 (compare_opt_names): New function.
4315 (arm_rewrite_selected_arch): Only strip out extensions that can be
4316 expressed through -mfpu. Sort the remaining extensions
4319 2017-06-08 Richard Earnshaw <rearnsha@arm.com>
4321 * config/arm/arm.h (arm_asm_auto_mfpu): Declare.
4322 (ASM_CPU_SPEC_FUNCTIONS): Add new rule asm_auto_mfpu.
4323 (ASM_CPU_SPEC): Use it if -mfpu is set to auto.
4324 * common/config/arm/arm-common.c (arm_asm_auto_mfpu): New function.
4326 2017-06-08 Tristan Gingold <gindold@adacore.com>
4329 * dwarf2out.c (dwarf2out_do_cfi_startproc): Only emit
4330 .cfi_personality or .cfi_lsda if the eh data format is dwarf2.
4332 2017-12-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4334 * config/sol2.h (SOLARIS_ATTRIBUTE_TABLE): Initialize new member
4335 of struct attribute_spec.
4337 2017-12-08 Julia Koval <julia.koval@intel.com>
4339 * config/i386/avx512vnniintrin.h (_mm512_dpwssds_epi32,
4340 _mm512_mask_dpwssds_epi32, _mm512_maskz_dpwssds_epi32): New intrinsics.
4341 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssds_epi32,
4342 _mm256_mask_dpwssds_epi32, _mm256_maskz_dpwssds_epi32,
4343 _mm_dpwssds_epi32, _mm_mask_dpwssds_epi32,
4344 _mm_maskz_dpwssds_epi32): Ditto.
4346 2017-12-08 Richard Biener <rguenther@suse.de>
4348 PR tree-optimization/81303
4349 * tree-vect-stmts.c (vect_is_simple_cond): For invariant
4350 conditions try to create a comparison vector type matching
4351 the data vector type.
4352 (vectorizable_condition): Adjust.
4353 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
4354 Leave invariant conditions alone in case we can vectorize those.
4356 2017-12-08 Julia Koval <julia.koval@intel.com>
4358 * config/i386/avx512vnniintrin.h (_mm512_dpwssd_epi32,
4359 _mm512_mask_dpwssd_epi32, _mm512_maskz_dpwssd_epi32): New intrinsics.
4360 * config/i386/avx512vnnivlintrin.h (_mm256_dpwssd_epi32,
4361 _mm256_mask_dpwssd_epi32, _mm256_maskz_dpwssd_epi32, _mm_dpwssd_epi32,
4362 _mm_mask_dpwssd_epi32, _mm_maskz_dpwssd_epi32): Ditto.
4364 2017-12-08 Julia Koval <julia.koval@intel.com>
4366 * config/i386/avx512vnniintrin.h (_mm512_dpbusds_epi32,
4367 _mm512_mask_dpbusds_epi32, _mm512_maskz_dpbusds_epi32): New.
4368 * config/i386/avx512vnnivlintrin.h (_mm256_dpbusds_epi32,
4369 _mm256_mask_dpbusds_epi32, _mm256_maskz_dpbusds_epi32,
4370 _mm_dpbusds_epi32, _mm_mask_dpbusds_epi32,
4371 _mm_maskz_dpbusds_epi32): New intrinsics.
4373 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4375 * config/nios2/nios2.md (ld<bhw_uns>io): Add splitter for memory
4377 (ld<bh>io_signed): Likewise.
4378 (st<bhw>io): Likewise.
4379 * config/nios2/predicates.md (ldstio_memory_operand): Allow
4380 SMALL_INT12 constant integer operand.
4382 2017-12-07 Sandra Loosemore <sandra@codesourcery.com>
4384 * config/nios2/nios2.c (nios2_symbolic_constant_allowed):
4386 (nios2_large_constant_allowed): ...this. Adjust uses.
4387 (nios2_plus_symbolic_constant_p): Rename to...
4388 (nios2_plus_large_constant_p): ...this. Adjust uses.
4389 (nios2_legitimate_address_p): Correct CONST_INT handling.
4390 (nios2_symbolic_memory_operand_p): Rename to...
4391 (nios2_large_constant_memory_operand_p): ...this. Adjust uses.
4392 (nios2_large_constant_p): Check for large constant integers too.
4393 (nios2_split_large_constant): Handle constant integers.
4394 (nios2_split_symbolic_memory_operand): Rename to...
4395 (nios2_split_large_constant_memory_operand): ...this. Adjust uses.
4396 (nios2_legitimize_constant_address): Handle constant integers.
4397 (r0rel_constant_p): Handle small constant integers.
4398 (nios2_print_operand_address): Handle r0-relative integer addresses.
4399 * config/nios2/nios2-protos.h: Adjust for renamed functions.
4400 * config/nios2/nios2.md: Adjust for renamed functions.
4402 2017-12-07 Andrew Waterman <andrew@sifive.com>
4404 * config/riscv/riscv.c (TARGET_ASM_SELECT_SECTION): New define.
4405 (TARGET_HAVE_SRODATA_SECTION): New define.
4406 (riscv_select_section): New function.
4408 2017-12-08 Joseph Myers <joseph@codesourcery.com>
4409 Alexander Monakov <amonakov@ispras.ru>
4410 Jakub Jelinek <jakub@redhat.com>
4413 * config/i386/i386.c (ix86_expand_rint): Handle flag_rounding_math.
4415 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4417 * config/aarch64/aarch64.c (aarch64_print_address_internal): Return
4418 a bool success value. Don't call output_operand_lossage here.
4419 (aarch64_print_ldpstp_address): Return a bool success value.
4420 (aarch64_print_operand_address): Call output_addr_const if
4421 aarch64_print_address_internal fails.
4422 (aarch64_print_operand): Don't assert that the mode is 16 bytes for
4423 'y'; call output_operand_lossage instead. Call output_operand_lossage
4424 if aarch64_print_ldpstp_address fails.
4426 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4428 * tree-vector-builder.h
4429 (tree_vector_builder::binary_encoded_nelts): Declare.
4430 * tree-vector-builder.c
4431 (tree_vector_builder::binary_encoded_nelts): New function.
4432 * fold-const.c (negate_expr_p): Likewise.
4433 (operand_equal_p, fold_checksum_tree): Likewise.
4434 * tree-loop-distribution.c (const_with_all_bytes_same): Likewise.
4435 * tree.c (integer_zerop, integer_onep, integer_all_onesp, real_zerop)
4436 (real_onep, real_minus_onep, add_expr, initializer_zerop): Likewise.
4437 (uniform_vector_p): Likewise.
4438 * varasm.c (const_hash_1, compare_constant): Likewise.
4439 * tree-ssa-ccp.c: Include tree-vector-builder.h.
4440 (valid_lattice_transition): Operate directly on the VECTOR_CST
4442 * ipa-icf.c: Include tree-vector-builder.h.
4443 (sem_variable::equals): Operate directly on the VECTOR_CST encoding.
4444 * print-tree.c (print_node): Print encoding of VECTOR_CSTs.
4446 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4448 * tree.c (build_vector): Delete.
4449 * tree.h (build_vector): Make static and move into the self-testing
4452 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4454 * vector-builder.h (vector_builder::derived): New const overload.
4455 (vector_builder::elt): New function.
4456 * tree-vector-builder.h (tree_vector_builder::type): New function.
4457 (tree_vector_builder::apply_step): Declare.
4458 * tree-vector-builder.c (tree_vector_builder::apply_step): New
4460 * gimple-fold.h (tree_vector_builder): Declare.
4461 (gimple_build_vector): Take a tree_vector_builder instead of a
4462 type and vector of elements.
4463 * gimple-fold.c (gimple_build_vector): Likewise.
4464 * tree-vect-loop.c (get_initial_def_for_reduction): Update call
4466 (get_initial_defs_for_reduction): Likewise.
4467 (vectorizable_induction): Likewise.
4469 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4471 * tree-vector-builder.h
4472 (tree_vector_builder::new_binary_operation): Declare.
4473 * tree-vector-builder.c
4474 (tree_vector_builder::new_binary_operation): New function.
4475 * fold-const.c (fold_relational_const): Use it.
4476 (const_binop): Likewise. Check that both input vectors have
4477 the same number of elements, thus excluding things like WIDEN_SUM.
4478 Check whether it is possible to operate directly on the encodings
4481 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4483 * fold-const.c (fold_negate_expr_1): Use tree_vector_builder and
4484 new_unary_operation, operating only on the encoded elements.
4485 (const_unop): Likewise.
4486 (exact_inverse): Likewise.
4487 (distributes_over_addition_p): New function.
4488 (const_binop): Use tree_vector_builder and new_unary_operation
4489 for combinations of VECTOR_CST and INTEGER_CST. Operate only
4490 on the encoded elements unless the encoding is strided and the
4491 operation does not distribute over addition.
4492 (fold_convert_const): Use tree_vector_builder and
4493 new_unary_operation. Operate only on the encoded elements
4494 for truncating integer conversions, or for non-stepped encodings.
4496 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4498 * config/sparc/sparc.c: Include tree-vector-builder.h.
4499 (sparc_fold_builtin): Use tree_vector_builder instead of build_vector.
4500 * expmed.c: Include tree-vector-builder.h.
4501 (make_tree): Use tree_vector_builder instead of build_vector.
4502 * fold-const.c: Include tree-vector-builder.h.
4503 (const_binop): Use tree_vector_builder instead of build_vector.
4504 (const_unop): Likewise.
4505 (native_interpret_vector): Likewise.
4506 (fold_vec_perm): Likewise.
4507 (fold_ternary_loc): Likewise.
4508 * gimple-fold.c: Include tree-vector-builder.h.
4509 (gimple_fold_stmt_to_constant_1): Use tree_vector_builder instead
4511 * tree-ssa-forwprop.c: Include tree-vector-builder.h.
4512 (simplify_vector_constructor): Use tree_vector_builder instead
4514 * tree-vect-generic.c: Include tree-vector-builder.h.
4515 (add_rshift): Use tree_vector_builder instead of build_vector.
4516 (expand_vector_divmod): Likewise.
4517 (optimize_vector_constructor): Likewise.
4518 * tree-vect-loop.c: Include tree-vector-builder.h.
4519 (vect_create_epilog_for_reduction): Use tree_vector_builder instead
4520 of build_vector. Explicitly use a stepped encoding for
4522 * tree-vect-slp.c: Include tree-vector-builder.h.
4523 (vect_get_constant_vectors): Use tree_vector_builder instead
4525 (vect_transform_slp_perm_load): Likewise.
4526 (vect_schedule_slp_instance): Likewise.
4527 * tree-vect-stmts.c: Include tree-vector-builder.h.
4528 (vectorizable_bswap): Use tree_vector_builder instead of build_vector.
4529 (vect_gen_perm_mask_any): Likewise.
4530 (vectorizable_call): Likewise. Explicitly use a stepped encoding.
4531 * tree.c: (build_vector_from_ctor): Use tree_vector_builder instead
4533 (build_vector_from_val): Likewise. Explicitly use a duplicate
4536 2017-12-07 Richard Sandiford <richard.sandiford@arm.com>
4538 * doc/generic.texi (VECTOR_CST): Describe new representation of
4540 * vector-builder.h: New file.
4541 * tree-vector-builder.h: Likewise.
4542 * tree-vector-builder.c: Likewise.
4543 * Makefile.in (OBJS): Add tree-vector-builder.o.
4544 * tree.def (VECTOR_CST): Update comment to refer to generic.texi.
4545 * tree-core.h (tree_base): Add a vector_cst field to the u union.
4546 (tree_vector): Change the number of elements to
4547 vector_cst_encoded_nelts.
4548 * tree.h (VECTOR_CST_NELTS): Redefine using TYPE_VECTOR_SUBPARTS.
4549 (VECTOR_CST_ELTS): Delete.
4550 (VECTOR_CST_ELT): Redefine using vector_cst_elt.
4551 (VECTOR_CST_LOG2_NPATTERNS, VECTOR_CST_NPATTERNS): New macros.
4552 (VECTOR_CST_NELTS_PER_PATTERN, VECTOR_CST_DUPLICATE_P): Likewise.
4553 (VECTOR_CST_STEPPED_P, VECTOR_CST_ENCODED_ELTS): Likewise.
4554 (VECTOR_CST_ENCODED_ELT): Likewise.
4555 (vector_cst_encoded_nelts): New function.
4556 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4557 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4558 (vector_cst_int_elt, vector_cst_elt): Declare.
4559 * tree.c: Include tree-vector-builder.h.
4560 (tree_code_size): Abort if passed VECTOR_CST.
4561 (tree_size): Update for new VECTOR_CST layout.
4562 (make_vector): Take the values of VECTOR_CST_LOG2_NPATTERNS and
4563 VECTOR_CST_NELTS_PER_PATTERN as arguments.
4564 (build_vector): Use tree_vector_builder.
4565 (vector_cst_int_elt, vector_cst_elt): New functions.
4566 (drop_tree_overflow): For VECTOR_CST, drop the TREE_OVERFLOW from the
4567 encoded elements and then create the vector in the canonical form.
4568 (check_vector_cst, check_vector_cst_duplicate, check_vector_cst_fill)
4569 (check_vector_cst_stepped, test_vector_cst_patterns): New functions.
4570 (tree_c_tests): Call test_vector_cst_patterns.
4571 * lto-streamer-out.c (DFS::DFS_write_tree_body): Handle the new
4573 (hash_tree): Likewise.
4574 * tree-streamer-out.c (write_ts_vector_tree_pointers): Likewise.
4575 (streamer_write_tree_header): Likewise.
4576 * tree-streamer-in.c (lto_input_ts_vector_tree_pointers): Likewise.
4577 (streamer_alloc_tree): Likewise. Update call to make_vector.
4578 * fold-const.c (fold_ternary_loc): Avoid using VECTOR_CST_ELTS.
4580 2017-12-07 Richard Sandiford <richard.sandiford@linaro.org>
4582 * selftest.h (ASSERT_TRUE_AT, ASSERT_FALSE_AT, ASSERT_EQ_AT)
4583 (ASSERT_NE, ASSERT_PRED1): Add underscores to local variable names
4584 * selftest-rtl.h (ASSERT_RTX_EQ, ASSERT_RTX_PTR_EQ): Likewise.
4586 2017-12-07 Bin Cheng <bin.cheng@arm.com>
4587 Richard Biener <rguenther@suse.de>
4589 PR tree-optimization/81303
4590 * Makefile.in (gimple-loop-interchange.o): New object file.
4591 * common.opt (floop-interchange): Reuse the option from graphite.
4592 * doc/invoke.texi (-floop-interchange): Ditto. New document for
4593 -floop-interchange and mention it for -O3.
4594 * opts.c (default_options_table): Enable -floop-interchange at -O3.
4595 * gimple-loop-interchange.cc: New file.
4596 * params.def (PARAM_LOOP_INTERCHANGE_MAX_NUM_STMTS): New parameter.
4597 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): New parameter.
4598 * passes.def (pass_linterchange): New pass.
4599 * timevar.def (TV_LINTERCHANGE): New time var.
4600 * tree-pass.h (make_pass_linterchange): New declaration.
4601 * tree-ssa-loop-ivcanon.c (create_canonical_iv): Change to external
4602 interchange. Record IV before/after increment in new parameters.
4603 * tree-ssa-loop-ivopts.h (create_canonical_iv): New declaration.
4604 * tree-vect-loop.c (vect_is_simple_reduction): Factor out reduction
4606 (check_reduction_path): ...New function here.
4607 * tree-vectorizer.h (check_reduction_path): New declaration.
4609 2017-12-07 Vladimir Makarov <vmakarov@redhat.com>
4612 PR rtl-optimization/80818
4613 * lra.c (add_regs_to_insn_regno_info): Make a hard reg in CLOBBER
4614 always early clobbered.
4615 * lra-lives.c (process_bb_lives): Check input hard regs for early
4616 clobbered non-operand hard reg.
4618 2017-12-07 Jakub Jelinek <jakub@redhat.com>
4621 * tree-cfg.c (verify_gimple_assign_binary): Don't require
4622 types_compatible_p, just that TYPE_MODE is the same.
4624 2017-12-07 Martin Sebor <msebor@redhat.com>
4627 * attribs.c (empty_attribute_table): Initialize new member of
4628 struct attribute_spec.
4629 (decl_attributes): Add argument. Handle mutually exclusive
4630 combinations of attributes.
4631 (selftests::test_attribute_exclusions): New function.
4632 (selftests::attribute_c_tests): Ditto.
4633 * attribs.h (decl_attributes): Add default argument.
4634 * selftest.h (attribute_c_tests): Declare.
4635 * selftest-run-tests.c (selftest::run_tests): Call attribute_c_tests.
4636 * tree-core.h (attribute_spec::exclusions, exclude): New type and
4638 * doc/extend.texi (Common Function Attributes): Update const and pure.
4639 * config/alpha/alpha.c (vms_attribute_table): Initialize new member
4640 of struct attribute_spec.
4641 * config/arc/arc.c (arc_attribute_table): Same.
4642 * config/arm/arm.c (arm_attribute_table): Same.
4643 * config/avr/avr.c ( avr_attribute_table): Same.
4644 * config/bfin/bfin.c (bfin_attribute_table): Same.
4645 * config/cr16/cr16.c (cr16_attribute_table): Same.
4646 * config/epiphany/epiphany.c (epiphany_attribute_table): Same.
4647 * config/h8300/h8300.c (h8300_attribute_table): Same.
4648 * config/i386/i386.c (ix86_attribute_table): Same.
4649 * config/ia64/ia64.c (ia64_attribute_table): Same.
4650 * config/m32c/m32c.c (m32c_attribute_table): Same.
4651 * config/m32r/m32r.c (m32r_attribute_table): Same.
4652 * config/m68k/m68k.c (m68k_attribute_table): Same.
4653 * config/mcore/mcore.c (mcore_attribute_table): Same.
4654 * config/microblaze/microblaze.c (microblaze_attribute_table): Same.
4655 * config/mips/mips.c (mips_attribute_table): Same.
4656 * config/msp430/msp430.c (msp430_attribute_table): Same.
4657 * config/nds32/nds32.c (nds32_attribute_table): Same.
4658 * config/nvptx/nvptx.c (nvptx_attribute_table): Same.
4659 * config/powerpcspe/powerpcspe.c (rs6000_attribute_table): Same.
4660 * config/rl78/rl78.c (rl78__attribute_table): Same.
4661 * config/rs6000/rs6000.c (rs6000_attribute_table): Same.
4662 * onfig/rx/rx.c (rx_attribute_table): Same.
4663 * config/s390/s390.c (s390_handle_vectorbool_attribute): Same.
4664 * config/sh/sh.c (sh_attribute_table): Same.
4665 * config/sparc/sparc.c (sparc_attribute_table): Same.
4666 * config/spu/spu.c (spu_attribute_table): Same.
4667 * config/stormy16/stormy16.c (xstormy16_attribute_table): Same.
4668 * config/v850/v850.c (v850_attribute_table): Same.
4669 * config/visium/visium.c (visium_attribute_table): Same.
4671 2017-12-07 Tamar Christina <tamar.christina@arm.com>
4674 * config/arm/arm.c (INCLUDE_STRING): Define.
4675 (arm_last_printed_arch_string, arm_last_printed_fpu_string): New.
4676 (arm_declare_function_name): Conservatively emit .arch, .arch_extensions
4679 2017-12-07 Michael Matz <matz@suse.de>
4681 Add unroll and jam pass
4683 * gimple-loop-jam.c: New file.
4684 * Makefile.in (OBJS): Add gimple-loop-jam.o.
4685 * common.opt (funroll-and-jam): New option.
4686 * opts.c (default_options_table): Add unroll-and-jam at -O3.
4687 * params.def (PARAM_UNROLL_JAM_MIN_PERCENT): New param.
4688 (PARAM_UNROLL_JAM_MAX_UNROLL): Ditto.
4689 * passes.def: Add pass_loop_jam.
4690 * timevar.def (TV_LOOP_JAM): Add.
4691 * tree-pass.h (make_pass_loop_jam): Declare.
4692 * cfgloop.c (flow_loop_tree_node_add): Add AT argument.
4693 * cfgloop.h (flow_loop_tree_node_add): Adjust declaration.
4694 * cfgloopmanip.c (duplicate_loop): Add AT argument, adjust call
4695 to flow_loop_tree_node_add.
4696 (duplicate_subloops, copy_loops_to): Append to sibling list.
4697 * cfgloopmanip.h: (duplicate_loop): Adjust declaration.
4698 * doc/invoke.texi (-funroll-and-jam): Document new option.
4699 (unroll-jam-min-percent, unroll-jam-max-unroll): Document new params.
4701 2017-12-07 Richard Biener <rguenther@suse.de>
4703 PR tree-optimization/83296
4704 PR tree-optimization/67769
4705 * tree-ssa-phiopt.c (conditional_replacement): Do not reset
4706 flow sensitive info in an unrelated BB.
4707 (value_replacement): Use reset_flow_sensitive_info.
4708 (minmax_replacement): Reset flow sensitive info on the def
4709 we move. Do not reset flow sensitive info in the whole BB
4710 we move the stmt to.
4711 (abs_replacement): Likewise.
4713 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4716 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4717 rs6000_cpu to the given -mcpu=, or to the default processor.
4719 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4721 * config/rs6000/rs6000.h (rs6000_cpu_attr): Delete.
4722 * config/rs6000/rs6000.c (rs6000_variable_issue_1): Use rs6000_tune
4723 instead of rs6000_cpu_attr.
4724 (rs6000_adjust_cost): Ditto.
4725 (is_microcoded_insn): Ditto.
4726 (rs6000_adjust_priority): Ditto.
4727 (rs6000_issue_rate): Ditto.
4728 (rs6000_use_sched_lookahead): Ditto.
4729 (rs6000_use_sched_lookahead_guard): Ditto.
4730 (rs6000_sched_reorder): Ditto.
4731 (force_new_group): Ditto.
4732 * config/rs6000/rs6000.md (cpu attribute): Ditto.
4733 (group_ending_nop): Ditto.
4735 2017-12-07 Segher Boessenkool <segher@kernel.crashing.org>
4737 * config/rs6000/rs6000.opt (rs6000_tune): New variable.
4738 * config/rs6000/rs6000.c (rs6000_option_override_internal): Also set
4739 rs6000_tune. Use rs6000_tune instead of rs6000_cpu where appropriate.
4740 (rs6000_loop_align): Use rs6000_tune instead of rs6000_cpu where
4742 (rs6000_reassociation_width): Ditto.
4743 (rs6000_emit_epilogue): Ditto.
4744 (rs6000_adjust_cost): Ditto.
4745 (is_microcoded_insn): Ditto.
4746 (is_cracked_insn): Ditto.
4747 (rs6000_adjust_priority): Ditto.
4748 (rs6000_sched_reorder): Ditto.
4749 (rs6000_sched_reorder2): Ditto.
4750 (insn_must_be_first_in_group): Ditto.
4751 (insn_must_be_last_in_group): Ditto.
4752 (rs6000_register_move_cost): Ditto.
4753 * config/rs6000/rs6000.h (rs6000_cpu_attr): Use rs6000_tune instead of
4756 2017-12-07 Julia Koval <julia.koval@intel.com>
4758 * config.gcc: Add vaesintrin.h.
4759 * config/i386/i386-builtin-types.def (V64QI_FTYPE_V64QI_V64QI):
4761 * config/i386/i386-builtin.def (__builtin_ia32_vaesdec_v16qi,
4762 __builtin_ia32_vaesdec_v32qi, __builtin_ia32_vaesdec_v64qi):
4764 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
4765 * config/i386/immintrin.h: Include vaesintrin.h.
4766 * config/i386/sse.md (vaesdec_<mode>): New pattern.
4767 * config/i386/vaesintrin.h (_mm256_aesdec_epi128, _mm512_aesdec_epi128,
4768 _mm_aesdec_epi128): New intrinsics.
4770 2017-12-06 David Malcolm <dmalcolm@redhat.com>
4772 * Makefile.in (C_COMMON_OBJS): Add c-family/c-spellcheck.o.
4773 * spellcheck-tree.c (find_closest_macro_cpp_cb): Move to
4774 c-family/c-spellcheck.cc.
4775 (best_macro_match::best_macro_match): Likewise.
4777 (struct edit_distance_traits<cpp_hashnode *>): Move to
4778 c-family/c-spellcheck.h.
4779 (class best_macro_match): Likewise.
4781 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4783 PR tree-optimization/83293
4784 * gimple-ssa-strength-reduction.c (insert_initializers): Use
4785 GSI_NEW_STMT instead of GSI_SAME_STMT in gsi_insert_after that
4786 might insert into empty bb.
4789 * match.pd ((T)(P + A) - (T)P -> (T) A): Split into separate
4790 simplify for plus with :c added, and pointer_plus without that.
4791 ((T)P - (T)(P + A) -> -(T) A): Likewise. If type is integral
4792 with undefined overflow and the conversion is not widening,
4793 perform negation in utype and only convert to type afterwards.
4794 ((T)(P + A) - (T)(P + B) -> (T)A - (T)B): Split into separate
4795 simplify for plus with :c added, and pointer_plus without that.
4796 If type is integral with undefined overflow and the conversion is
4797 not widening, perform minus in utype and only convert to type
4798 afterwards. Move the last pointer_diff_expr simplify into the
4801 2017-12-06 Martin Sebor <msebor@redhat.com>
4803 PR tree-optimization/82646
4804 * builtins.c (maybe_emit_chk_warning): Use size as the bound for
4805 strncpy, not maxlen.
4807 2017-12-06 Martin Sebor <msebor@redhat.com>
4809 * doc/invoke.texi (-Wstringop-truncation): Mention attribute
4812 PR tree-optimization/83075
4813 * tree-ssa-strlen.c (handle_builtin_stxncpy): Avoid assuming
4814 strncat/strncpy don't change length of source string.
4816 2017-12-06 Eric Botcazou <ebotcazou@adacore.com>
4819 2017-11-29 Martin Aberg <maberg@gaisler.com>
4821 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
4822 to prevent b2bst errata sequence.
4823 (sqrtdf2_fix): Likewise.
4825 2017-12-06 Jakub Jelinek <jakub@redhat.com>
4827 PR tree-optimization/81945
4828 * cfgloop.h (FOR_EACH_LOOP_FN): Use FN instead of hardcoding fn.
4829 * tree-cfg.c (move_sese_region_to_fn): If any of the loops moved
4830 to dest_cfun has orig_loop_num set, either remap it to the new
4831 loop number if the loop got moved too, or clear it.
4833 2017-12-05 Steve Ellcey <sellcey@cavium.com>
4835 * config/aarch64/thunderx2-t99.md (thunderx2t99_branch): Add trap
4837 (thunderx2t99_nothing): New insn reservation.
4838 (thunderx2t99_mrs): New insn reservation.
4839 (thunderx2t99_multiple): New insn reservation.
4840 (thunderx2t99_alu_basi): Add bfx to reservation.
4841 (thunderx2t99_fp_cmp): Add fccmps and fccmpd to reservation.
4843 2017-12-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
4846 * config/arm/arm.md (probe_stack) : Use the 'o' constraint.
4848 2017-12-05 Bin Cheng <bin.cheng@arm.com>
4850 * tree-ssa-dce.c (simple_dce_from_worklist): Move and rename from
4851 tree-ssa-pre.c::remove_dead_inserted_code.
4852 * tree-ssa-dce.h: New file.
4853 * tree-ssa-pre.c (tree-ssa-dce.h): Include new header file.
4854 (remove_dead_inserted_code): Move and rename to function
4855 tree-ssa-dce.c::simple_dce_from_worklist.
4856 (pass_pre::execute): Update use.
4858 2017-12-05 Richard Biener <rguenther@suse.de>
4860 PR tree-optimization/83277
4861 * graphite-isl-ast-to-gimple.c (should_copy_to_new_region): Make sure
4862 to code-gen liveout vars.
4864 2017-12-05 Richard Sandiford <richard.sandiford@linaro.org>
4866 * config/aarch64/aarch64-simd.md (aarch64_simd_bsldi_internal)
4867 (aarch64_simd_bsldi_alt): Check REG_P before GP_REGNUM_P.
4868 (aarch64_cm<optab>di, aarch64_cmtstdi): Add leading "&&" to
4871 2017-12-05 Max Filippov <jcmvbkbc@gmail.com>
4873 * config/xtensa/xtensa.c (xtensa_asan_shadow_offset): New
4875 (TARGET_ASAN_SHADOW_OFFSET): New macro definition.
4876 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): Set to 1 if
4879 2017-12-05 Richard Biener <rguenther@suse.de>
4881 * timevar.def (TV_TREE_RECIP, TV_TREE_SINCOS, TV_TREE_WIDEN_MUL): Add.
4882 * tree-ssa-math-opts.c (pass_data_cse_reciprocal): Use TV_TREE_RECIP.
4883 (pass_data_cse_sincos): Use TV_TREE_SINCOS.
4884 (pass_data_optimize_widening_mul): Use TV_TREE_WIDEN_MUL.
4886 2017-12-05 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4888 * dbxout.c (dbxout_block): Grow buf to 30 bytes.
4890 2017-12-05 Martin Liska <mliska@suse.cz>
4891 Jakub Jelinek <jakub@redhat.com>
4893 * doc/invoke.texi: Document the options.
4894 * flag-types.h (enum sanitize_code): Add
4895 SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4896 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Add handling
4897 of SANITIZE_POINTER_COMPARE and SANITIZE_POINTER_SUBTRACT.
4898 * opts.c: Define new sanitizer options.
4899 * sanitizer.def (BUILT_IN_ASAN_POINTER_COMPARE): Likewise.
4900 (BUILT_IN_ASAN_POINTER_SUBTRACT): Likewise.
4902 2017-12-05 Julia Koval <julia.koval@intel.com>
4904 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VNNI_SET,
4905 OPTION_MASK_ISA_AVX512VNNI_UNSET): New.
4906 (ix86_handle_option): Handle -mavx512vnni.
4907 * config/i386/cpuid.h (bit_AVX512VNNI): New bit.
4908 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
4909 * config/i386/i386-c (__AVX512VNNI__): New.
4910 * config/i386/i386.c (ix86_target_string): Handle new option.
4911 (ix86_valid_target_attribute_inner_p): Handle new option.
4912 * config/i386/i386.h (TARGET_AVX512VNNI, TARGET_AVX512VNNI_P): New.
4913 * config/i386/i386.opt (mavx512vnni): New option.
4915 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
4918 * athlon.md: Disable for generic.
4919 * haswell.md: Enable for generic.
4920 * i386.c (ix86_sched_init_global): Add core hooks for generic.
4921 * x86-tune-sched.c (ix86_issue_rate): Increase issue rate for generic
4923 (ix86_adjust_cost): Move generic to haswell path.
4925 2017-12-04 Eric Botcazou <ebotcazou@adacore.com>
4927 * config/sparc/sparc.c (sparc_do_work_around_errata): Use mem_ref
4928 instead of MEM_P in a couple more places. Fix formatting issues.
4930 2017-12-04 Jim Wilson <jimw@sifive.com>
4932 * config/riscv/riscv.c (riscv_for_each_saved_reg): Use GP_REG_LAST
4933 instead of GP_REG_LAST-1.
4934 (riscv_adjust_libcall_cfi_prologue): Likewise.
4935 (riscv_adjust_libcall_cri_epilogue): Likewise.
4936 * config/riscv/riscv.h (CALL_USED_REGISTERS): Change a6 to t6 in
4939 2017-12-04 Luis Machado <luis.machado@linaro.org>
4941 * ipa-pure-const.c (check_decl): Add missing newline.
4942 (state_from_flags): Likewise.
4944 2017-12-04 Jeff Law <law@redhat.com>
4946 PR tree-optimizatin/78496
4947 * gimple-ssa-evrp-analyze.h
4948 (evrp_range_analyzer::get_vr_values): Simplify.
4949 * gimple-ssa-evrp-analyze.c: Corresponding changes.
4950 * tree-ssa-dom.c: Include alloc-pool.h, tree-vrp.h, vr-values.h
4951 and gimple-ssa-evrp-analyze.h.
4952 (dom_opt_dom_walker class): Add evrp_range_analyzer member.
4953 (simplify_stmt_for_jump_threading): Copy a blob of code from
4954 tree-vrp.c to use ranges to simplify statements.
4955 (dom_opt_dom_walker::before_dom_children): Call
4956 evrp_range_analyzer::{enter,record_ranges_from_stmt} methods.
4957 (dom_opt_dom_walker::after_dom_children): Similarly for
4958 evrp_range_analyzer::leave.
4959 (dom_opt_dom_walker::optimize_stmt): Use EVRP ranges to optimize
4962 * gimple-ssa-evrp-analyze.c
4963 (evrp_range_analyzer::extract_range_from_stmt): Always use
4964 vr_values::update_value_range so preexisting range info is
4965 medged with new range info, even if the new range is VR_VARYING.
4967 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4969 * combine.c: Adjust comment.
4970 (use_crosses_set_p): Delete.
4971 (can_combine_p): Use modified_between_p instead of use_crosses_set_p.
4972 (try_combine): Ditto.
4974 2017-12-04 Richard Biener <rguenther@suse.de>
4976 PR tree-optimization/83255
4977 * graphite-isl-ast-to-gimple.c (translate_isl_ast_node_for):
4978 Re-add zero-iteration check.
4980 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4982 PR rtl-optimization/83245
4983 * lra.c (collect_non_operand_hard_regs): Treat clobbers of non-operand
4984 hard registers as earlyclobber, also if not in an asm.
4986 2017-12-04 Segher Boessenkool <segher@kernel.crashing.org>
4990 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
4993 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
4994 rs6000_cpu based on cpu_index, not tune_index.
4996 2017-12-04 Richard Biener <rguenther@suse.de>
4998 PR tree-optimization/83238
4999 * graphite-scop-detection.c (scop_detection::merge_sese): Make
5000 code match comment, rejecting invalid SESE regions.
5002 2017-12-03 John David Anglin <danglin@gcc.gnu.org>
5004 * config/pa/pa.c (pa_legitimate_address_p): For scaled indexing,
5005 require base operand is a REG_POINTER prior to reload on targets
5006 with non-equivalent space registers.
5008 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
5010 * ipa-cp.c (ipcp_lattice<valtype>::print): Update dumping.
5011 (update_specialized_profile): Fix updating of counts.
5012 (perhaps_add_new_callers): Likewise.
5014 2017-12-01 Jan Hubicka <hubicka@ucw.cz>
5017 * x86-tune.def: Remove obsolette FIXMEs.
5018 (X86_TUNE_PARTIAL_FLAG_REG_STALL): Disable for generic
5019 (X86_TUNE_FUSE_CMP_AND_BRANCH_32, X86_TUNE_FUSE_CMP_AND_BRANCH_64,
5020 X86_TUNE_FUSE_CMP_AND_BRANCH_SOFLAGS, X86_TUNE_FUSE_ALU_AND_BRANCH):
5022 (X86_TUNE_PAD_RETURNS): Disable for generic.
5024 2017-12-02 Jakub Jelinek <jakub@redhat.com>
5026 PR tree-optimization/83170
5027 PR tree-optimization/83241
5028 * gimple-ssa-store-merging.c
5029 (imm_store_chain_info::try_coalesce_bswap): Update vuse field from
5030 gimple_vuse (ins_stmt) in case it has changed.
5031 (imm_store_chain_info::output_merged_store): Likewise.
5033 * tree-chkp.c (chkp_compute_bounds_for_assignment): Handle
5037 * tree-cfg.c (pass_warn_function_return::execute): Handle
5038 __builtin_ubsan_handle_missing_return like __builtin_unreachable
5039 with BUILTINS_LOCATION.
5043 * expr.c (get_inner_reference): If DECL_MODE of a non-bitfield
5044 is BLKmode for vector field with vector raw mode, use TYPE_MODE
5045 instead of DECL_MODE.
5047 * config/i386/i386-protos.h (standard_sse_constant_opcode): Change
5048 last argument to rtx pointer.
5049 * config/i386/i386.c (standard_sse_constant_opcode): Replace X argument
5050 with OPERANDS. For AVX+ 128-bit VEX encoded instructions over 256-bit
5051 or 512-bit. If setting EXT_REX_SSE_REG_P, use EVEX encoded insn
5052 depending on the chosen ISAs.
5053 * config/i386/i386.md (*movxi_internal_avx512f, *movoi_internal_avx,
5054 *movti_internal, *movdi_internal, *movsi_internal, *movtf_internal,
5055 *movdf_internal, *movsf_internal): Adjust standard_sse_constant_opcode
5057 * config/i386/sse.md (mov<mode>_internal): Likewise.
5058 * config/i386/mmx.md (*mov<mode>_internal): Likewise.
5060 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5062 * doc/invoke.texi (-dp): Say that instruction cost is printed as well.
5064 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5066 * config/rs6000/rs6000-protos.h (emit_fusion_addis): Remove last two
5067 parameters from prototype.
5068 * config/rs6000/rs6000.c (emit_fusion_addis): Remove last two
5069 parameters. Don't print a comment.
5070 (emit_fusion_gpr_load): Adjust.
5071 (emit_fusion_load_store): Adjust.
5072 * config/rs6000/rs6000.md (*fusion_p9_<mode>_constant): Adjust.
5073 * config/rs6000/vsx.md (two peepholes): Print the "vector load fusion"
5074 comment on the second line.
5076 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5079 * config/rs6000/rs6000.c (rs6000_option_override_internal): Set
5080 rs6000_cpu based on cpu_index, not tune_index.
5082 2017-12-01 Segher Boessenkool <segher@kernel.crashing.org>
5084 * final.c (output_asm_name): Print insn_cost. Shorten output. Print
5085 which_alternative instead of which_alternative + 1.
5086 (output_asm_insn): Print an extra tab if the template is short.
5088 2017-12-01 Jim Wilson <jimw@sifive.com>
5090 * common.opt (use_gnu_debug_info_extensions): Delete DWARF_DEBUG from
5092 * config/vx-common.h (DWARF_DEBUGGING_INFO): Delete undef.
5093 * doc/tm.texi.in (PREFERRED_DEBUGGING_TYPE): Delete DWARF_DEBUG
5095 * doc/tm.texi: Regenerate.
5097 2017-12-01 Michael Meissner <meissner@linux.vnet.ibm.com>
5100 * config/rs6000/rs6000.c (rs6000_address_for_fpconvert): Check for
5101 whether we can allocate pseudos before trying to fix an address.
5102 * config/rs6000/rs6000.md (float_<mode>si2_hw): Make sure the
5103 memory address is indexed or indirect.
5104 (floatuns_<mode>si2_hw2): Likewise.
5106 2017-12-01 Jason Merrill <jason@redhat.com>
5108 * Makefile.in (TAGS): Add c-family/*.cc.
5110 2017-12-01 Wilco Dijkstra <wdijkstr@arm.com>
5112 * config/aarch64/aarch64.md (call_insn): Use %c rather than %a.
5113 (call_value_insn): Likewise.
5114 (sibcall_insn): Likewise.
5115 (sibcall_value_insn): Likewise.
5116 (movsi_aarch64): Likewise.
5117 (movdi_aarch64): Likewise.
5118 (add_losym_): Likewise.
5119 (ldr_got_small_): Likewise.
5120 (ldr_got_small_sidi): Likewise.
5121 (ldr_got_small_28k_): Likewise.
5122 (ldr_got_small_28k_sidi): Likewise.
5123 * config/aarch64/aarch64.c (aarch64_print_address_internal):
5124 Move output_addr_const to symbolic case. Add error check.
5126 2017-12-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5128 * config/s390/predicates.md (plus16_Q_operand): New predicate.
5129 * config/s390/s390.md: Disable MVC merging peephole if it would
5130 disable operand forwarding.
5131 (new peephole2): Split MVCs if it would turn them into up to 2
5134 2017-12-01 Richard Biener <rguenther@suse.de>
5136 PR tree-optimization/83232
5137 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Fix
5138 detection of same access. Instead of breaking the group here
5139 do not consider the duplicate. Add comment explaining real fix.
5141 2017-12-01 Jonathan Wakely <jwakely@redhat.com>
5143 * doc/md.texi (Insn Splitting): Fix "central flowgraph" typo.
5145 2017-12-01 Sudakshina Das <sudi.das@arm.com>
5147 * config/arm/vfp.md (*movhf_vfp_fp16): Add conds attribute.
5149 2017-12-01 Jakub Jelinek <jakub@redhat.com>
5151 * function.h (struct function): Remove cilk_frame_decl,
5152 is_cilk_function and calls_cilk_spawn fields.
5153 * tree-inline.h (struct copy_body_data): Remove remap_var_for_cilk
5155 * omp-simd-clone.c (simd_clone_clauses_extract): Don't clear
5156 cilk_elemental field.
5157 * cgraph.h (struct cgraph_simd_clone): Remove cilk_elemental field.
5158 * target.def: Adjust comment.
5159 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
5160 Don't test cilk_elemental.
5162 PR tree-optimization/83233
5163 * gimple-ssa-store-merging.c (nop_stats, bswap_stats): Use
5164 bswap_stat name for the struct.
5167 * tree.h (SWITCH_BREAK_LABEL_P): Define.
5168 * gimplify.c (collect_fallthrough_labels): Handle GIMPLE_BIND
5169 starting with a GIMPLE_SWITCH and ending with GIMPLE_LABEL with
5170 SWITCH_BREAK_LABEL_P set on the label.
5171 (gimplify_switch_expr): Set SWITCH_BREAK_LABEL_P on the label
5172 added for default case if it was missing and not all cases covered.
5173 Wrap GIMPLE_SWITCH and the switch_body_seq into a GIMPLE_BIND if
5174 switch_body_seq ends with a GIMPLE_LABEL with SWITCH_BREAK_LABEL_P
5176 * tree-chrec.c (evolution_function_is_univariate_p): Add return true;
5177 to avoid -Wimplicit-fallthrough warning.
5178 * config/i386/i386.c (ix86_expand_special_args_builtin): Add
5179 FALLTHRU comment to avoid -Wimplicit-fallthrough warning.
5181 PR tree-optimization/83221
5182 * tree-ssa-reassoc.c (sort_by_operand_rank): Shift bb_rank
5184 (init_reassoc): Formatting fix.
5187 * tree-cfg.c (group_case_labels_stmt): Don't optimize away
5188 C++ FE implicitly added __builtin_unreachable () until -Wreturn-type
5192 * tree-cfg.c: Include asan.h.
5193 (gimple_seq_unreachable_p): Return false for -fsanitize=unreachable.
5195 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5197 * config/i386/i386.md: Fix AVX512 register width in AVX512 instruction.
5199 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5201 * config/i386/i386.c (standard_sse_constant_opcode): Fix wrong form for
5202 vpcmpeqd instruction.
5204 2017-12-01 Sergey Shalnov <Sergey.Shalnov@intel.com>
5206 * config/i386/i386.c (standard_sse_constant_opcode): Fix
5207 registers type for 128bit mode.
5209 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5211 * spellcheck-tree.c (test_find_closest_identifier): Use ; instead
5213 * gengtype-state.c (read_state_pair): Likewise.
5214 * gimple-fold.c (gimple_fold_builtin_string_compare): Likewise.
5215 * sel-sched-dump.c (dump_insn_rtx_1): Likewise.
5216 * ipa-cp.c (intersect_aggregates_with_edge): Likewise.
5217 * ifcvt.c (noce_try_store_flag_constants): Likewise.
5218 * tree-ssa-ccp.c (ccp_finalize): Likewise.
5219 * omp-grid.c (grid_process_kernel_body_copy): Likewise.
5220 * builtins.c (fold_builtin_3): Likewise.
5221 * graphite-scop-detection.c
5222 (scop_detection::stmt_has_simple_data_refs_p): Likewise.
5223 * hsa-gen.c (hsa_function_representation::hsa_function_representation):
5226 2017-12-01 Maxim Ostapenko <m.ostapenko@samsung.com>
5229 * asan.c (asan_protect_global): Add new ignore_decl_rtl_set_p
5230 parameter. Return true if ignore_decl_rtl_set_p is true and other
5231 conditions are satisfied.
5232 * asan.h (asan_protect_global): Add new parameter.
5233 * varasm.c (categorize_decl_for_section): Pass true as second parameter
5234 to asan_protect_global calls.
5236 2017-11-30 Jim Wilson <jimw@sifive.com>
5238 * doc/invoke.texi (RISC-V Options): Delete nonexistent -mmemcpy and
5239 -mno-memcpy options. For -mplt, -mfdiv, -mdiv, -msave-restore, and
5240 -mstrict-align, add info on default value. Delete redundant lines for
5241 -mabi. Add missing -mexplicit-relocs docs.
5243 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5245 * config/arc/arc.md (trap): New pattern.
5247 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5249 * config/arc/arc.c (hwloop_optimize): Prevent the last ZOL
5250 instruction to end into a delay slot.
5251 * config/arc/arc.md (cond_delay_insn): Check if the instruction
5252 can be placed into a delay slot against reg_note.
5254 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5256 * config/arc/arc.c (hwloop_optimize): Update hw-loop's end/start
5257 labels number of usages.
5259 2017-11-30 Claudiu Zissulescu <claziss@synopsys.com>
5261 * config/arc/arc.c (arc_cannot_substitue_mem_equiv_p): New
5263 (TARGET_CANNOT_SUBSTITUTE_MEM_EQUIV_P): Define.
5265 2017-11-30 Jakub Jelinek <jakub@redhat.com>
5268 * internal-fn.c (expand_mul_overflow): Optimize unsigned
5269 multiplication by power of 2 constant into two shifts + comparison.
5271 2017-11-30 Jan Hubicka <hubicka@ucw.cz>
5274 * config/i386/x86-tune-costs.h (generic_cost): Revise for modern CPUs.
5276 2017-11-30 Richard Biener <rguenther@suse.de>
5278 PR tree-optimization/83202
5279 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Add
5280 allow_peel argument and guard peeling.
5281 (canonicalize_loop_induction_variables): Likewise.
5282 (canonicalize_induction_variables): Pass false.
5283 (tree_unroll_loops_completely_1): Pass unroll_outer to disallow
5284 peeling from cunrolli.
5286 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5288 * combine.c (try_combine): Print a message to dump file whenever
5289 I0, I1, or I2 cannot be combined into I3.
5291 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5293 PR rtl-optimization/83156
5294 PR rtl-optimization/82621
5295 * combine.c (try_combine): Don't split an I2 if one of the dests is
5296 set again before I3. Allow unused dests.
5298 2017-11-29 Segher Boessenkool <segher@kernel.crashing.org>
5300 * config/rs6000/rs6000.md (*add<mode>3_carry_in_internal2): New.
5302 2017-11-29 Vladimir Makarov <vmakarov@redhat.com>
5304 PR rtl-optimization/80818
5305 * lra.c (collect_non_operand_hard_regs): New arg insn. Pass it
5306 recursively. Use insn code for clobber.
5307 (lra_set_insn_recog_data): Pass the new arg to
5308 collect_non_operand_hard_regs.
5309 (add_regs_to_insn_regno_info): Pass insn instead of uid. Use insn
5311 (lra_update_insn_regno_info): Pass insn to
5312 add_regs_to_insn_regno_info.
5314 2017-11-29 Jim Wilson <jimw@sifive.com>
5315 Andrew Waterman <andrew@sifive.com>
5317 * config/riscv/riscv.c (SINGLE_SHIFT_COST): New.
5318 (riscv_rtx_costs): Case ZERO_EXTRACT, match new pattern, and return
5319 SINGLE_SHIFT_COST. Case LT and ZERO_EXTEND, likewise. Case ASHIFT,
5320 use SINGLE_SHIFT_COST.
5321 * config/riscv/riscv.md (lshrsi3_zero_extend_1): New.
5322 (lshrsi3_zero_extend_2, lshrsi3_zero_extend_3): New.
5324 2017-11-29 Julia Koval <julia.koval@intel.com>
5326 * config/i386/avx512vbmi2intrin.h (_mm512_shldv_epi16,
5327 _mm512_mask_shldv_epi16, _mm512_maskz_shldv_epi16, _mm512_shldv_epi32,
5328 _mm512_mask_shldv_epi32, _mm512_maskz_shldv_epi32, _mm512_shldv_epi64,
5329 _mm512_mask_shldv_epi64, _mm512_maskz_shldv_epi64): New intrinsics.
5330 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldv_epi16,
5331 _mm256_mask_shldv_epi16, _mm256_maskz_shldv_epi16, _mm256_shldv_epi32,
5332 _mm256_mask_shldv_epi32, _mm256_maskz_shldv_epi32, _mm256_shldv_epi64,
5333 _mm256_mask_shldv_epi64, _mm256_maskz_shldv_epi64, _mm_shldv_epi16,
5334 _mm_mask_shldv_epi16, _mm_maskz_shldv_epi16, _mm_shldv_epi32,
5335 _mm_mask_shldv_epi32, _mm_maskz_shldv_epi32, _mm_shldv_epi64,
5336 _mm_mask_shldv_epi64, _mm_maskz_shldv_epi64): Ditto.
5337 * config/i386/i386-builtin.def (__builtin_ia32_vpshldv_v32hi,
5338 __builtin_ia32_vpshldv_v32hi_mask, __builtin_ia32_vpshldv_v32hi_maskz,
5339 __builtin_ia32_vpshldv_v16hi, __builtin_ia32_vpshldv_v16hi_mask,
5340 __builtin_ia32_vpshldv_v16hi_maskz, __builtin_ia32_vpshldv_v8hi,
5341 __builtin_ia32_vpshldv_v8hi_mask, __builtin_ia32_vpshldv_v8hi_maskz,
5342 __builtin_ia32_vpshldv_v16si, __builtin_ia32_vpshldv_v16si_mask,
5343 __builtin_ia32_vpshldv_v16si_maskz, __builtin_ia32_vpshldv_v8si,
5344 __builtin_ia32_vpshldv_v8si_mask, __builtin_ia32_vpshldv_v8si_maskz,
5345 __builtin_ia32_vpshldv_v4si, __builtin_ia32_vpshldv_v4si_mask,
5346 __builtin_ia32_vpshldv_v4si_maskz, __builtin_ia32_vpshldv_v8di,
5347 __builtin_ia32_vpshldv_v8di_mask, __builtin_ia32_vpshldv_v8di_maskz,
5348 __builtin_ia32_vpshldv_v4di, __builtin_ia32_vpshldv_v4di_mask,
5349 __builtin_ia32_vpshldv_v4di_maskz, __builtin_ia32_vpshldv_v2di,
5350 __builtin_ia32_vpshldv_v2di_mask,
5351 __builtin_ia32_vpshldv_v2di_maskz): New builtins.
5352 * config/i386/sse.md (vpshldv_<mode>, vpshldv_<mode>_mask,
5353 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): New patterns.
5355 2017-11-29 Julia Koval <julia.koval@intel.com>
5357 * config/i386/avx512vbmi2intrin.h (_mm512_shrdv_epi16,
5358 _mm512_mask_shrdv_epi16, _mm512_maskz_shrdv_epi16, _mm512_shrdv_epi32,
5359 _mm512_mask_shrdv_epi32, _mm512_maskz_shrdv_epi32, _mm512_shrdv_epi64,
5360 _mm512_mask_shrdv_epi64, _mm512_maskz_shrdv_epi64): New intrinsics.
5361 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdv_epi16,
5362 _mm256_mask_shrdv_epi16, _mm256_maskz_shrdv_epi16, _mm256_shrdv_epi32,
5363 _mm256_mask_shrdv_epi32, _mm256_maskz_shrdv_epi32, _mm256_shrdv_epi64,
5364 _mm256_mask_shrdv_epi64, _mm256_maskz_shrdv_epi64, _mm_shrdv_epi16,
5365 _mm_mask_shrdv_epi16, _mm_maskz_shrdv_epi16, _mm_shrdv_epi32,
5366 _mm_mask_shrdv_epi32, _mm_maskz_shrdv_epi32, _mm_shrdv_epi64,
5367 _mm_mask_shrdv_epi64, _mm_maskz_shrdv_epi64): Ditto.
5368 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_V32HI,
5369 V32HI_FTYPE_V32HI_V32HI_V32HI_INT, V16HI_FTYPE_V16HI_V16HI_V16HI_INT,
5370 V8HI_FTYPE_V8HI_V8HI_V8HI_INT, V8SI_FTYPE_V8SI_V8SI_V8SI_INT,
5371 V4SI_FTYPE_V4SI_V4SI_V4SI_INT, V8DI_FTYPE_V8DI_V8DI_V8DI,
5372 V8DI_FTYPE_V8DI_V8DI_V8DI_INT, V4DI_FTYPE_V4DI_V4DI_V4DI_INT,
5373 V16SI_FTYPE_V16SI_V16SI_V16SI, V16SI_FTYPE_V16SI_V16SI_V16SI_INT,
5374 V2DI_FTYPE_V2DI_V2DI_V2DI_INT): New types.
5375 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5376 * config/i386/sse.md (vpshrdv_<mode>, vpshrdv_<mode>_mask,
5377 vpshrdv_<mode>_maskz, vpshrdv_<mode>_maskz_1): New pattern.
5379 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5381 * config/sparc/sparc.c (sparc_do_work_around_errata): Treat the
5382 movsi_pic_gotdata_op instruction as a load for the UT699 errata
5385 2017-11-29 Martin Aberg <maberg@gaisler.com>
5387 * config/sparc/sparc.md (divdf3_fix): Add NOP and adjust length
5388 to prevent b2bst errata sequence.
5389 (sqrtdf2_fix): Likewise.
5391 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5393 * config/sparc/sparc.c (fpop_reg_depend_p): New function.
5394 (div_sqrt_insn_p): New function.
5395 (sparc_do_work_around_errata): Insert NOP instructions to
5396 prevent sequences that could trigger the TN-0013 errata for
5397 certain LEON3 processors.
5398 (pass_work_around_errata::gate): Also test sparc_fix_lost_divsqrt.
5399 (sparc_option_override): Set sparc_fix_lost_divsqrt appropriately.
5400 * config/sparc/sparc.md (fix_lost_divsqrt): New attribute.
5401 (in_branch_delay): Prevent div and sqrt in delay slot if
5403 * config/sparc/sparc.opt (sparc_fix_lost_divsqrt): New variable.
5405 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5407 * config/sparc/sparc.c (atomic_insn_p): New function.
5408 (sparc_do_work_around_errata): Insert NOP instructions to
5409 prevent sequences that could trigger the TN-0010 errata for
5411 * config/sparc/sync.md (atomic_compare_and_swap_leon3_1): Make
5412 instruction referable in atomic_insns_p.
5414 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5416 * config/sparc/sync.md (swapsi): 16-byte align if sparc_fix_gr712rc.
5417 (atomic_compare_and_swap_leon3_1): Likewise.
5420 2017-11-29 Daniel Cederman <cederman@gaisler.com>
5422 * config/sparc/sparc.c (fpop_insn_p): New function.
5423 (sparc_do_work_around_errata): Insert NOP instructions to
5424 prevent sequences that could trigger the TN-0012 errata for
5426 (pass_work_around_errata::gate): Also test sparc_fix_gr712rc.
5427 * config/sparc/sparc.md (fix_gr712rc): New attribute.
5428 (in_branch_annul_delay): Prevent floating-point instructions
5429 in delay slot of annulled integer branch.
5431 2017-11-29 Richard Biener <rguenther@suse.de>
5433 PR tree-optimization/83202
5434 * tree-vect-slp.c (scalar_stmts_set_t): New typedef.
5436 (vect_analyze_slp_cost_1): Add visited set, do not account SLP
5437 nodes vectorized to the same stmts multiple times.
5438 (vect_analyze_slp_cost): Allocate a visited set and pass it down.
5439 (vect_analyze_slp_instance): Adjust.
5440 (scalar_stmts_to_slp_tree_map_t): New typedef.
5441 (vect_schedule_slp_instance): Add a map recording the SLP node
5442 representing the vectorized stmts for a set of scalar stmts.
5443 Avoid code-generating redundancies.
5444 (vect_schedule_slp): Allocate map and pass it down.
5446 2017-11-29 Nathan Sidwell <nathan@acm.org>
5449 * tree.c (build_complex_type): Fix canonicalization. Only fill in
5452 2017-11-29 Wilco Dijkstra <wdijkstr@arm.com>
5454 * config/aarch64/aarch64.c (aarch64_print_operand): Add new
5455 cases for printing LDP/STP memory addresses.
5456 (aarch64_print_address_internal): Renamed from
5457 aarch64_print_operand_address, added parameter, add Pmode check.
5458 (aarch64_print_ldpstp_address): New function for LDP/STP addresses.
5459 (aarch64_print_operand_address): Indirect to
5460 aarch64_print_address_internal.
5461 * config/aarch64/aarch64-simd.md (store_pair_lanes): Use new
5462 'y' operand output specifier.
5464 2017-11-29 Jakub Jelinek <jakub@redhat.com>
5467 * tree.c (build_simple_mem_ref_loc): Handle
5468 get_addr_base_and_unit_offset returning a MEM_REF.
5471 * rtlanal.c (seq_cost): For non-single_set insns try to use insn_cost.
5474 * config/i386/sse.md (vec_concatv2di): Remove * from (=Yr,0,*rm)
5477 2017-11-28 David Malcolm <dmalcolm@redhat.com>
5479 * diagnostic-show-locus.c (layout::print_trailing_fixits): Handle
5481 (layout::move_to_column): Likewise.
5483 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5486 * tree.c (block_may_fallthru): Return false if SWITCH_ALL_CASES_P
5487 is set on SWITCH_EXPR and !block_may_fallthru (SWITCH_BODY ()).
5489 2017-11-28 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5490 Martin Jambor <mjambor@suse.cz>
5493 * tree.h (expr_type_first_operand_type_p): Declare
5494 * tree.c (expr_type_first_operand_type_p): New function.
5495 * ipa-prop.h (ipa_get_type): Allow i to be out of bounds.
5496 (ipa_value_from_jfunc): Adjust declaration.
5497 * ipa-cp.c (ipa_get_jf_pass_through_result): New parameter RES_TYPE.
5498 Use it as result type for arithmetics, unless it is NULL in which case
5499 be more conservative.
5500 (ipa_value_from_jfunc): New parameter PARM_TYPE, pass it to
5501 ipa_get_jf_pass_through_result.
5502 (propagate_vals_across_pass_through): Likewise.
5503 (propagate_scalar_across_jump_function): New parameter PARM_TYPE, pass
5504 is to propagate_vals_across_pass_through.
5505 (propagate_constants_across_call): Pass PARM_TYPE to
5506 propagate_scalar_across_jump_function.
5507 (find_more_scalar_values_for_callers_subset): Pass parameter type to
5508 ipa_value_from_jfunc.
5509 (cgraph_edge_brings_all_scalars_for_node): Likewise.
5510 * ipa-fnsummary.c (evaluate_properties_for_edge): Renamed parms_info
5511 to caller_parms_info, pass parameter type to ipa_value_from_jfunc.
5512 * ipa-prop.c (try_make_edge_direct_simple_call): New parameter
5513 target_type, pass it to ipa_value_from_jfunc.
5514 (update_indirect_edges_after_inlining): Pass parameter type to
5515 try_make_edge_direct_simple_call.
5517 2017-11-28 Jeff Law <law@redhat.com>
5519 * gimple-ssa-evrp-analyze.c
5520 (evrp_range_analyzer::record_ranges_from_phis): Only use SCEV to
5521 refine ranges if scev_initialized_p returns true.
5522 * vr-values.c (vr_values::extract_range_from_phi_node): Likewise.
5524 2017-11-28 Julia Koval <julia.koval@intel.com>
5526 * config/i386/avx512vbmi2intrin.h (_mm512_shrdi_epi16,
5527 _mm512_mask_shrdi_epi16, _mm512_maskz_shrdi_epi16, _mm512_shrdi_epi32,
5528 _mm512_mask_shrdi_epi32, _mm512_maskz_shrdi_epi32, _mm512_shrdi_epi64,
5529 _mm512_mask_shrdi_epi64, _mm512_maskz_shrdi_epi64): New intrinsics.
5530 * config/i386/avx512vbmi2vlintrin.h (_mm256_shrdi_epi16,
5531 _mm256_mask_shrdi_epi16, _mm256_maskz_shrdi_epi16,
5532 _mm256_mask_shrdi_epi32, _mm256_maskz_shrdi_epi32, _mm256_shrdi_epi32,
5533 _mm256_mask_shrdi_epi64, _mm256_maskz_shrdi_epi64, _mm256_shrdi_epi64,
5534 _mm_mask_shrdi_epi16, _mm_maskz_shrdi_epi16, _mm_shrdi_epi16,
5535 _mm_mask_shrdi_epi32, _mm_maskz_shrdi_epi32, _mm_shrdi_epi32,
5536 _mm_mask_shrdi_epi64, _mm_maskz_shrdi_epi64, _mm_shrdi_epi64): Ditto.
5537 * config/i386/i386-builtin.def (__builtin_ia32_vpshrd_v32hi,
5538 __builtin_ia32_vpshrd_v32hi_mask, __builtin_ia32_vpshrd_v16hi,
5539 __builtin_ia32_vpshrd_v16hi_mask, __builtin_ia32_vpshrd_v8hi,
5540 __builtin_ia32_vpshrd_v8hi_mask, __builtin_ia32_vpshrd_v16si,
5541 __builtin_ia32_vpshrd_v16si_mask, __builtin_ia32_vpshrd_v8si,
5542 __builtin_ia32_vpshrd_v8si_mask, __builtin_ia32_vpshrd_v4si,
5543 __builtin_ia32_vpshrd_v4si_mask, __builtin_ia32_vpshrd_v8di,
5544 __builtin_ia32_vpshrd_v8di_mask, __builtin_ia32_vpshrd_v4di,
5545 __builtin_ia32_vpshrd_v4di_mask, __builtin_ia32_vpshrd_v2di,
5546 __builtin_ia32_vpshrd_v2di_mask): New builtins.
5547 * config/i386/sse.md (vpshrd_<mode><mask_name>): New pattern.
5549 2017-11-28 Julia Koval <julia.koval@intel.com>
5551 * config/i386/avx512vbmi2intrin.h (_mm512_shldi_epi16,
5552 _mm512_mask_shldi_epi16, _mm512_maskz_shldi_epi16, _mm512_shldi_epi32,
5553 _mm512_mask_shldi_epi32, _mm512_maskz_shldi_epi32, _mm512_shldi_epi64,
5554 _mm512_mask_shldi_epi64, _mm512_maskz_shldi_epi64): New intrinsics.
5555 * config/i386/avx512vbmi2vlintrin.h (_mm256_shldi_epi16,
5556 _mm256_mask_shldi_epi16, _mm256_maskz_shldi_epi16,
5557 _mm256_mask_shldi_epi32, _mm256_maskz_shldi_epi32, _mm256_shldi_epi32,
5558 _mm256_mask_shldi_epi64, _mm256_maskz_shldi_epi64, _mm256_shldi_epi64,
5559 _mm_mask_shldi_epi16, _mm_maskz_shldi_epi16, _mm_shldi_epi16,
5560 _mm_mask_shldi_epi32, _mm_maskz_shldi_epi32, _mm_shldi_epi32,
5561 _mm_mask_shldi_epi64, _mm_maskz_shldi_epi64, _mm_shldi_epi64): Ditto.
5562 * config/i386/i386-builtin-types.def (V32HI_FTYPE_V32HI_V32HI_INT,
5563 V32HI_FTYPE_V32HI_V32HI_INT_V32HI_INT, V16SI_FTYPE_V16SI_V16SI_INT,
5564 V16SI_FTYPE_V16SI_V16SI_INT_V16SI_INT,
5565 V8DI_FTYPE_V8DI_V8DI_INT_V8DI_INT, V8SI_FTYPE_V8SI_V8SI_INT_V8SI_INT,
5566 V16HI_FTYPE_V16HI_V16HI_INT_V16HI_INT,
5567 V4DI_FTYPE_V4DI_V4DI_INT_V4DI_INT,
5568 V8HI_FTYPE_V8HI_V8HI_INT_V8HI_INT,
5569 V4SI_FTYPE_V4SI_V4SI_INT_V4SI_INT,
5570 V2DI_FTYPE_V2DI_V2DI_INT_V2DI_INT): New types.
5571 * config/i386/i386-builtin.def (__builtin_ia32_vpshld_v32hi,
5572 __builtin_ia32_vpshld_v32hi_mask, __builtin_ia32_vpshld_v16hi,
5573 __builtin_ia32_vpshld_v16hi_mask, __builtin_ia32_vpshld_v8hi,
5574 __builtin_ia32_vpshld_v8hi_mask, __builtin_ia32_vpshld_v16si,
5575 __builtin_ia32_vpshld_v16si_mask, __builtin_ia32_vpshld_v8si,
5576 __builtin_ia32_vpshld_v8si_mask, __builtin_ia32_vpshld_v4si,
5577 __builtin_ia32_vpshld_v4si_mask, __builtin_ia32_vpshld_v8di,
5578 __builtin_ia32_vpshld_v8di_mask, __builtin_ia32_vpshld_v4di,
5579 __builtin_ia32_vpshld_v4di_mask, __builtin_ia32_vpshld_v2di,
5580 __builtin_ia32_vpshld_v2di_mask): New builtins.
5581 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
5582 * config/i386/sse.md (vpshld_<mode><mask_name>): New pattern.
5584 2017-11-28 Richard Biener <rguenther@suse.de>
5586 PR tree-optimization/80776
5587 * gimple-ssa-evrp-analyze.h (evrp_range_analyzer::set_ssa_range_info):
5589 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::set_ssa_range_info):
5591 (evrp_range_analyzer::record_ranges_from_incoming_edges):
5592 If the incoming edge is an effective fallthru because the other
5593 edge only reaches a __builtin_unreachable () then record ranges
5594 derived from the controlling condition in SSA info.
5595 (evrp_range_analyzer::record_ranges_from_phis): Use set_ssa_range_info.
5596 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
5598 2017-11-28 Olivier Hainque <hainque@adacore.com>
5600 * Makefile.in (SELFTEST_FLAGS): Use nul instead of /dev/null
5601 on mingw build hosts.
5603 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5606 * dbxout.c (dbx_block_with_cold_children): Fix function comment.
5608 2017-11-28 Richard Biener <rguenther@suse.de>
5611 * gimple-fold.c (gimple_fold_builtin_memory_op): For aggregate
5612 copies generated from memcpy use a character array as reference
5615 2017-11-28 Julia Koval <julia.koval@intel.com>
5616 Sebastian Peryt <sebastian.peryt@intel.com>
5618 * Makefile.in (cilkplus.def, cilk-builtins.def, c-family/cilk.o,
5619 c-family/c-cilkplus.o, c-family/array-notation-common.o,
5620 cilk-common.o, cilk.h, cilk-common.c): Remove.
5622 (BT_FN_INT_PTR_PTR_PTR_FTYPE_BT_INT_BT_PTR_BT_PTR_BT_PTR): Remove.
5623 * builtins.c (is_builtin_name): Remove cilkplus condition.
5624 (BUILT_IN_CILK_DETACH, BUILT_IN_CILK_POP_FRAME): Remove.
5625 * builtins.def (DEF_CILK_BUILTIN_STUB, DEF_CILKPLUS_BUILTIN,
5626 cilk-builtins.def, cilkplus.def): Remove.
5627 * cif-code.def (CILK_SPAWN): Remove.
5628 * cilk-builtins.def: Delete.
5629 * cilk-common.c: Ditto.
5631 * cilkplus.def: Ditto.
5632 * config/darwin.h (fcilkplus): Delete.
5633 * cppbuiltin.c: Ditto.
5634 * doc/extend.texi: Remove cilkplus doc.
5635 * doc/generic.texi: Ditto.
5636 * doc/invoke.texi: Ditto.
5637 * doc/passes.texi: Ditto.
5638 * gcc.c (fcilkplus): Remove.
5639 * gengtype.c (cilk.h): Remove.
5640 * gimple-pretty-print.c (dump_gimple_omp_for): Remove cilkplus
5642 * gimple.h (GF_OMP_FOR_KIND_CILKFOR, GF_OMP_FOR_KIND_CILKSIMD):
5644 * gimplify.c (gimplify_return_expr, maybe_fold_stmt,
5646 is_gimple_stmt, gimplify_modify_expr, gimplify_scan_omp_clauses,
5647 gimplify_adjust_omp_clauses, gimplify_omp_for, gimplify_expr): Remove
5648 cilkplus conditions.
5649 * ipa-fnsummary.c (ipa_dump_fn_summary, compute_fn_summary,
5650 inline_read_section): Ditto.
5651 * ipa-inline-analysis.c (cilk.h): Remove.
5652 * ira.c (ira_setup_eliminable_regset): Remove cilkplus support.
5653 * lto-wrapper.c (merge_and_complain, append_compiler_options,
5654 append_linker_options): Remove condition for fcilkplus.
5655 * lto/lto-lang.c (cilk.h): Remove.
5656 (lto_init): Remove condition for fcilkplus.
5657 * omp-expand.c (expand_cilk_for_call): Delete.
5658 (expand_omp_taskreg, expand_omp_for_static_chunk,
5659 expand_omp_for): Remove cilkplus
5661 (expand_cilk_for): Delete.
5662 * omp-general.c (omp_extract_for_data): Remove cilkplus support.
5663 * omp-low.c (scan_sharing_clauses, create_omp_child_function,
5664 execute_lower_omp, diagnose_sb_0): Ditto.
5665 * omp-simd-clone.c (simd_clone_clauses_extract): Ditto.
5666 * tree-core.h (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5667 * tree-nested.c: Ditto.
5668 * tree-pretty-print.c (dump_omp_clause): Remove cilkplus support.
5669 (dump_generic_node): Ditto.
5670 * tree.c (OMP_CLAUSE__CILK_FOR_COUNT_): Delete.
5671 * tree.def (cilk_simd, cilk_for, cilk_spawn_stmt, cilk_sync_stmt):
5673 * tree.h (CILK_SPAWN_FN, EXPR_CILK_SPAWN): Delete.
5675 2017-11-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5677 * config/aarch64/aarch64.md (div<mode>3): Change check to TARGET_FLOAT.
5678 * config/aarch64/aarch64.c (aarch64_emit_approx_div): Add early exit
5679 for vector mode and !TARGET_SIMD.
5681 2017-11-28 Jakub Jelinek <jakub@redhat.com>
5683 * tree.def (SWITCH_EXPR): Change from 3 operand to 2 operand tree.
5685 * tree.h (SWITCH_LABELS): Remove.
5686 * gimplify.c (gimplify_switch_expr): Don't test SWITCH_LABELS,
5687 assert SWITCH_BODY is non-NULL.
5688 * tree-pretty-print.c (dump_generic_node): Remove SWITCH_LABELS
5690 * tree.c (block_may_fallthru): Always return true; for SWITCH_EXPR.
5692 PR tree-optimization/80788
5693 * match.pd (X +- C1 CMP C2 -> X CMP C2 -+ C1): If res
5694 has TREE_OVERFLOW set, call drop_tree_overflow.
5696 2017-11-28 Richard Biener <rguenther@suse.de>
5698 PR tree-optimization/83158
5699 * tree-vrp.c (intersect_ranges): Prefer ~[0, 0] in a few more cases.
5701 2017-11-28 Segher Boessenkool <segher@kernel.crashing.org>
5704 * config/rs6000/rs6000.c (rs6000_rtx_costs): Do not handle
5705 TARGET_ISEL && !TARGET_MFCRF differently. Simplify code.
5707 2017-11-27 Segher Boessenkool <segher@kernel.crashing.org>
5709 * config/rs6000/rs6000.md (<code><GPR:mode><GPR2:mode>2_isel): Change
5710 LT/GT/LTU/GTU to LE/GE/LEU/GEU where possible.
5712 2017-11-27 Michael Meissner <meissner@linux.vnet.ibm.com>
5715 * varasm.c (compare_constant): Take the mode of the constants into
5716 account when comparing floating point constants.
5718 2017-11-27 Gerald Pfeifer <gerald@pfeifer.com>
5720 * hash-set.h (DEFINE_DEBUG_HASH_SET): Remove static qualifier
5721 from explicit instantiation of debug_helper.
5722 * vec.h (DEFINE_DEBUG_VEC): Ditto.
5724 2017-11-27 Richard Biener <rguenther@suse.de>
5726 * gimple-fold.c (gimple_fold_builtin_memory_op): Remove dead code,
5729 2017-11-27 Richard Biener <rguenther@suse.de>
5731 * tree.c (wide_int_to_tree): Free discarded INTEGER_CST.
5732 (type_hash_canon): Also clear int_cst_hash_table entry for
5734 (build_nonstandard_integer_type): Hash all TYPE_MAX_VALUEs.
5736 2017-11-27 Tamar Christina <tamar.christina@arm.com>
5738 * doc/extend.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5739 * doc/invoke.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5740 * doc/sourcebuild.texi: Add -A suffix (ARMv8*-A, ARMv7-A).
5742 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5744 * hash-map.h (gt_cleare_cache): Avoid UB.
5746 2017-11-27 Eric Botcazou <ebotcazou@adacore.com>
5748 * cfgloop.h (struct loop): Document usage of USHRT_MAX for unroll.
5749 * loop-unroll.c (decide_unroll_constant_iterations): Implement it.
5750 (decide_unroll_runtime_iterations): Likewise.
5751 (decide_unroll_stupid): Likewise.
5753 2017-11-27 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
5756 * config/i386/i386.md: Add a loop with incssp.
5758 2017-11-27 Martin Jambor <mjambor@suse.cz>
5760 PR tree-optimization/81248
5761 * tree-sra.c (splice_param_accesses): Remove size check.
5762 (decide_one_param_reduction): Fix size check.
5763 * gimple-pretty-print.c (dump_profile): Silence warning.
5764 * params.def (PARAM_IPA_SRA_PTR_GROWTH_FACTOR): Adjust description.
5766 2017-11-27 Jakub Jelinek <jakub@redhat.com>
5769 * dbxout.c (lastlineno): New variable.
5770 (dbx_debug_hooks): Use dbxout_switch_text_section as
5771 switch_text_section debug hook.
5772 (dbxout_function_end): Switch to current_function_section
5773 rather than function_section. If crtl->has_bb_partition,
5774 output just one N_FUN, depending on in_cold_section_p.
5775 (dbxout_source_line): Remember last lineno in lastlineno.
5776 (dbxout_switch_text_section): New function.
5777 (dbxout_function_decl): Adjust dbxout_block caller.
5778 (dbx_block_with_cold_children): New function.
5779 (dbxout_block): Return true if any LBRAC/RBRAC have been
5780 emitted. Use dbx_block_with_cold_children at depth == 0
5781 in second partition. Add PARENT_BLOCKNUM argument, pass
5782 it optionally adjusted to children. Output LBRAC/RBRAC
5783 around recursive call only if the block is in the current
5784 partition, if not and anything was output, emit empty
5786 * final.c (final_scan_insn): Compute cold_function_name
5787 before calling switch_text_section debug hook. Call
5788 that hook even if dwarf2out_do_frame if not emitting
5792 * varasm.c (bss_initializer_p): Return true for DECL_COMMON
5793 TREE_READONLY decls.
5795 2017-11-27 Markus Trippelsdorf <markus@trippelsdorf.de>
5797 PR rtl-optimization/82488
5798 * expr.c (fixup_args_size_notes): Avoid signed integer overflow.
5800 2017-11-26 Julia Koval <julia.koval@intel.com>
5802 * config/i386/i386.c (processor_target_table): Add skylake_cost for
5804 * config/i386/x86-tune-costs.h (skylake_memcpy, skylake_memset,
5807 2017-11-26 Julia Koval <julia.koval@intel.com>
5809 * config/i386/driver-i386.c (host_detect_local_cpu):
5810 Detect skylake-avx512.
5812 2017-11-26 Julia Koval <julia.koval@intel.com>
5814 * config.gcc: Add -march=cannonlake.
5815 * config/i386/driver-i386.c (host_detect_local_cpu): Detect cannonlake.
5816 * config/i386/i386-c.c (ix86_target_macros_internal): Handle cannonlake.
5817 * config/i386/i386.c (processor_costs): Add m_CANNONLAKE.
5818 (PTA_CANNONLAKE): New.
5819 (processor_target_table): Add cannonlake.
5820 (ix86_option_override_internal): Ditto.
5821 (fold_builtin_cpu): Ditto.
5822 (get_builtin_code_for_version): Handle cannonlake.
5823 (M_INTEL_COREI7_CANNONLAKE): New.
5824 * config/i386/i386.h (TARGET_CANNONLAKE, PROCESSOR_CANNONLAKE): New.
5825 * doc/invoke.texi: Add -march=cannonlake.
5827 2017-11-14 Boris Kolpackov <boris@codesynthesis.com>
5829 * plugin.c (add_new_plugin): Use platform-specific library extensions.
5830 (try_init_one_plugin): Alternative implementation for MinGW.
5831 * Makefile.in (plugin_implib): New.
5832 (gengtype-lex.c): Fix broken AIX workaround.
5833 * configure: Regenerate.
5834 * doc/plugins.texi: Document support for MinGW.
5836 2017-11-25 Jakub Jelinek <jakub@redhat.com>
5838 PR rtl-optimization/81553
5839 * combine.c (simplify_if_then_else): In (if_then_else COND (OP Z C1) Z)
5840 to (OP Z (mult COND (C1 * STORE_FLAG_VALUE))) optimization, if OP
5841 is a shift where C1 has different mode than the whole shift, use C1's
5842 mode for MULT rather than the shift's mode.
5845 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Don't fold
5846 builtins not enabled in the currently selected ISA.
5848 2017-11-24 Jackson Woodruff <jackson.woodruff@arm.com>
5850 PR tree-optimization/71026
5851 * tree-ssa-math-opts (is_division_by_square, is_square_of): New.
5852 (insert_reciprocals): Change to insert reciprocals before a division
5853 by a square and to insert the square of a reciprocal.
5854 (execute_cse_reciprocals_1): Change to consider division by a square.
5855 (register_division_in): Add importance parameter.
5857 2017-11-24 Richard Biener <rguenther@suse.de>
5859 PR tree-optimization/82402
5860 * tree-vect-loop-manip.c (create_lcssa_for_virtual_phi): Properly
5861 set SSA_NAME_OCCURS_IN_ABNORMAL_PHI.
5863 2017-11-24 Marc Glisse <marc.glisse@inria.fr>
5865 * match.pd (0-ptr): New transformation.
5867 2017-11-24 Jan Hubicka <hubicka@ucw.cz>
5870 * ipa-inline.c (inline_small_functions): Set current badnes correctly
5871 when skipping checking.
5873 2017-11-24 Richard Biener <rguenther@suse.de>
5875 PR tree-optimization/83128
5876 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Handle STRING_CSTs.
5877 (vn_reference_lookup_3): Likewise.
5879 2017-11-24 Jakub Jelinek <jakub@redhat.com>
5882 * ubsan.c (ubsan_type_descriptor): Use pp_unsigned_wide_integer
5883 instead of pp_printf with HOST_WIDE_INT_PRINT_DEC. Avoid calling
5886 * tree-object-size.c (pass_through_call): Use gimple_call_return_flags
5887 ERF_RETURN*ARG* for builtins other than BUILT_IN_ASSUME_ALIGNED,
5888 check for the latter with gimple_call_builtin_p. Do not handle
5889 BUILT_IN_STPNCPY_CHK which is not a pass through call.
5891 2017-11-24 Christophe Lyon <christophe.lyon@linaro.org>
5893 * config/arm/arm_neon.h: Fix pragma GCC push_options before
5896 2017-11-23 Julia Koval <julia.koval@intel.com>
5898 * config/i386/avx512vbmi2intrin.h (_mm512_mask_expand_epi8,
5899 _mm512_maskz_expand_epi8, _mm512_mask_expandloadu_epi8,
5900 _mm512_maskz_expandloadu_epi8, _mm512_mask_expand_epi16,
5901 _mm512_maskz_expand_epi16, _mm512_mask_expandloadu_epi16,
5902 _mm512_maskz_expandloadu_epi16): New intrinsics.
5903 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_expand_epi8,
5904 _mm_maskz_expand_epi8, _mm_mask_expandloadu_epi8,
5905 _mm_maskz_expandloadu_epi8, _mm_mask_expand_epi16,
5906 _mm_maskz_expand_epi16, _mm_mask_expandloadu_epi16,
5907 _mm_maskz_expandloadu_epi16, _mm256_mask_expand_epi16,
5908 _mm256_maskz_expand_epi16, _mm256_mask_expandloadu_epi16,
5909 _mm256_maskz_expandloadu_epi16, _mm256_mask_expand_epi8,
5910 _mm256_maskz_expand_epi8, _mm256_mask_expandloadu_epi8,
5911 _mm256_maskz_expandloadu_epi8): New intrinsics.
5912 * config/i386/i386-builtin-types.def (V64QI_FTYPE_PCV64QI_V64QI_UDI,
5913 V32HI_FTYPE_PCV32HI_V32HI_USI, V32QI_FTYPE_PCV32QI_V32QI_USI,
5914 V16HI_FTYPE_PCV16HI_V16HI_UHI, V16QI_FTYPE_PCV16QI_V16QI_UHI,
5915 V8HI_FTYPE_PCV8HI_V8HI_UQI): New types.
5916 * config/i386/i386.c (ix86_expand_special_args_builtin): Use new types.
5917 * config/i386/sse.md (VI248_VLBW): New iterator.
5918 (expand<mode>_mask, expand<mode>_maskz): New patterns.
5920 2017-11-23 Julia Koval <julia.koval@intel.com>
5922 * config.gcc (avx512vbmi2intrin.h, avx512vbmi2vlintrin): New headers.
5923 * config/i386/avx512vbmi2intrin.h (_mm512_mask_compress_epi8,
5924 _mm512_maskz_compress_epi8, _mm512_mask_compressstoreu_epi8,
5925 _mm512_mask_compress_epi16, _mm512_maskz_compress_epi16,
5926 _mm512_mask_compressstoreu_epi16): New.
5927 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_compress_epi8,
5928 _mm_maskz_compress_epi8, _mm256_mask_compressstoreu_epi16,
5929 _mm_mask_compress_epi16, _mm_maskz_compress_epi16,
5930 _mm256_mask_compress_epi16, _mm256_maskz_compress_epi16,
5931 _mm_mask_compressstoreu_epi8, _mm_mask_compressstoreu_epi16,
5932 _mm256_mask_compress_epi8, _mm256_maskz_compress_epi8,
5933 _mm256_mask_compressstoreu_epi8): New.
5934 * config/i386/i386-builtin-types.def (VOID_FTYPE_PV64QI_V64QI_UDI,
5935 VOID_FTYPE_PV32HI_V32HI_USI, VOID_FTYPE_PV32QI_V32QI_USI,
5936 VOID_FTYPE_PV16QI_V16QI_UHI, VOID_FTYPE_PV16HI_V16HI_UHI,
5937 VOID_FTYPE_PV8HI_V8HI_UQI): New types.
5938 * config/i386/i386-builtin.def (__builtin_ia32_compressqi512_mask,
5939 __builtin_ia32_compresshi512_mask, __builtin_ia32_compressqi256_mask,
5940 __builtin_ia32_compressqi128_mask, __builtin_ia32_compresshi256_mask,
5941 __builtin_ia32_compresshi128_mask,
5942 __builtin_ia32_compressstoreuqi512_mask,
5943 __builtin_ia32_compressstoreuhi512_mask,
5944 __builtin_ia32_compressstoreuqi256_mask,
5945 __builtin_ia32_compressstoreuqi128_mask,
5946 __builtin_ia32_compressstoreuhi256_mask,
5947 __builtin_ia32_compressstoreuhi128_mask): New builtins.
5948 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Create special args
5950 (ix86_expand_special_args_builtin): Handle new types.
5951 (s4fma_expand): Handle new builtin array.
5952 * config/i386/immintrin.h: Include new headers.
5953 * config/i386/sse.md (VI12_AVX512VLBW): New iterator.
5954 (compress<mode>_mask, compressstore<mode>_mask): New patterns.
5956 2017-11-23 Jakub Jelinek <jakub@redhat.com>
5958 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Formatting
5959 fixes. Declare temp and g variables at the top in order to avoid
5960 {} in most of the cases.
5962 2017-11-23 Marc Glisse <marc.glisse@inria.fr>
5964 * match.pd (ptr-0): New transformation.
5966 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
5968 * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
5969 (struct addr_mode_cost_table): New.
5970 (struct tune_params): Add field addr_mode_costs.
5971 * config/arm/arm.c (generic_addr_mode_costs): New.
5972 (arm_slowmul_tune): Initialise addr_mode_costs field.
5973 (arm_fastmul_tune): Likewise.
5974 (arm_strongarm_tune): Likewise.
5975 (arm_xscale_tune): Likewise.
5976 (arm_9e_tune): Likewise.
5977 (arm_marvell_pj4_tune): Likewise.
5978 (arm_v6t2_tune): Likewise.
5979 (arm_cortex_tune): Likewise.
5980 (arm_cortex_a8_tune): Likewise.
5981 (arm_cortex_a7_tune): Likewise.
5982 (arm_cortex_a15_tune): Likewise.
5983 (arm_cortex_a35_tune): Likewise.
5984 (arm_cortex_a53_tune): Likewise.
5985 (arm_cortex_a57_tune): Likewise.
5986 (arm_exynosm1_tune): Likewise.
5987 (arm_xgene1_tune): Likewise.
5988 (arm_cortex_a5_tune): Likewise.
5989 (arm_cortex_a9_tune): Likewise.
5990 (arm_cortex_a12_tune): Likewise.
5991 (arm_cortex_a73_tune): Likewise.
5992 (arm_v7m_tune): Likewise.
5993 (arm_cortex_m7_tune): Likewise.
5994 (arm_v6m_tune): Likewise.
5995 (arm_fa726te_tune): Likewise.
5996 (arm_mem_costs): Use table lookup to calculate cost of addressing mode.
5998 2017-11-23 Charles Baylis <charles.baylis@linaro.org>
6000 * config/arm/arm.c (arm_mem_costs): New function.
6001 (arm_rtx_costs_internal): Use arm_mem_costs.
6003 2017-11-23 Mark Wielaard <mark@klomp.org>
6005 * dwarf2out.c (init_sections_and_labels): Use generation to create
6006 unique ranges_section_label and ranges_base_label. Return generation.
6007 (output_rnglists): Add generation argument. Use generation to create
6008 unique ranges labels.
6009 (dwarf2out_finish): Get generation from init_sections_and_labels
6010 and pass generation to output_rnglists.
6012 2017-11-23 Mike Stump <mikestump@comcast.net>
6013 Eric Botcazou <ebotcazou@adacore.com>
6015 * doc/generic.texi (ANNOTATE_EXPR): Document 3rd operand.
6016 * cfgloop.h (struct loop): Add unroll field.
6017 * function.h (struct function): Add has_unroll bitfield.
6018 * gimplify.c (gimple_boolify) <ANNOTATE_EXPR>: Deal with unroll kind.
6019 (gimplify_expr) <ANNOTATE_EXPR>: Propagate 3rd operand.
6020 * loop-init.c (pass_loop2::gate): Return true if cfun->has_unroll.
6021 (pass_rtl_unroll_loops::gate): Likewise.
6022 * loop-unroll.c (decide_unrolling): Tweak note message. Skip loops
6023 for which loop->unroll==1.
6024 (decide_unroll_constant_iterations): Use note for consistency and
6025 take loop->unroll into account. Return early if loop->unroll is set.
6026 Fix thinko in existing test.
6027 (decide_unroll_runtime_iterations): Use note for consistency and
6028 take loop->unroll into account.
6029 (decide_unroll_stupid): Likewise.
6030 * lto-streamer-in.c (input_cfg): Read loop->unroll.
6031 * lto-streamer-out.c (output_cfg): Write loop->unroll.
6032 * tree-cfg.c (replace_loop_annotate_in_block) <annot_expr_unroll_kind>:
6034 (replace_loop_annotate) <annot_expr_unroll_kind>: Likewise.
6035 (print_loop): Print loop->unroll if set.
6036 * tree-core.h (enum annot_expr_kind): Add annot_expr_unroll_kind.
6037 * tree-inline.c (copy_loops): Copy unroll and set cfun->has_unroll.
6038 * tree-pretty-print.c (dump_generic_node) <annot_expr_unroll_kind>:
6040 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Bail out if
6041 loop->unroll is set and smaller than the trip count. Otherwise bypass
6042 entirely the heuristics if loop->unroll is set. Remove dead note.
6043 Fix off-by-one bug in other note.
6044 (try_peel_loop): Bail out if loop->unroll is set. Fix formatting.
6045 (tree_unroll_loops_completely_1): Force unrolling if loop->unroll
6047 (tree_unroll_loops_completely): Make static.
6048 (pass_complete_unroll::execute): Use correct type for variable.
6049 (pass_complete_unrolli::execute): Fix formatting.
6050 * tree.def (ANNOTATE_EXPR): Add 3rd operand.
6052 2017-11-23 Sergey Shalnov <Sergey.Shalnov@intel.com>
6054 * config/i386/i386.h (TARGET_PREFER_AVX256): Also
6055 enable when TARGET_PREFER_AVX128 is set.
6057 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6059 * ipa-profile.c (ipa_propagate_frequency_1): Use count instead of
6061 * cgraph.c (cgraph_edge::dump_edge_flags): Dump sreal frequencies.
6062 (cgraph_edge::maybe_hot_p): Use sreal frequencies.
6064 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6066 * ipa-fnsummary.c (record_modified_bb_info): Use sreal
6067 frequencies. Fix estimation of aggregate parameters.
6069 2017-11-23 Jan Hubicka <hubicka@ucw.cz>
6071 * cgraphclones.c (cgraph_node::create_clone): Fix updating of profile
6074 2017-11-23 Tom de Vries <tom@codesourcery.com>
6076 * config/sh/sh.h (ASM_OUTPUT_ADDR_VEC_ELT): Wrap in "do {} while (0)".
6078 2017-11-23 Tom de Vries <tom@codesourcery.com>
6080 * config/ft32/ft32.h (ASM_OUTPUT_ADDR_VEC_ELT): Remove semicolon after
6083 2017-11-23 Oleg Endo <olegendo@gcc.gnu.org>
6086 * config/sh/sh.md (udivsi3, divsi3, sibcall_value_pcrel,
6087 sibcall_value_pcrel_fdpic): Use local variable instead of
6089 (calli_tbr_rel): Add missing operand 2.
6090 (call_valuei_tbr_rel): Add missing operand 3.
6092 2017-11-23 Jakub Jelinek <jakub@redhat.com>
6095 * expr.c (expand_assignment): For CONCAT to_rtx, complex type from and
6096 bitpos/bitsize covering the whole destination, use store_expr only if
6097 the complex mode is the same. Otherwise, use expand_normal and if
6098 it returns CONCAT, subreg each part separately instead of trying to
6099 subreg the whole result.
6101 2017-11-23 Richard Biener <rguenther@suse.de>
6103 PR tree-optimization/23094
6104 * tree-ssa-sccvn.c (vuse_ssa_val): Handle VN_TOP when we
6105 come here from walking over backedges in the first iteration.
6106 (vn_reference_lookup_3): Skip clobbers that store the same value.
6108 2017-11-23 Richard Biener <rguenther@suse.de>
6110 PR tree-optimization/81403
6111 * tree-ssa-pre.c (get_representative_for): Add parameter specifying
6112 a block we need a leader relative to.
6113 (phi_translate_1): For nary processing require a leader from
6114 get_representative_for given we run expression simplification
6115 using match-and-simplify. Remove previous fix.
6117 2017-11-22 Jeff Law <law@redhat.com>
6119 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
6120 Use new method allocate_value_range rather than accessing the
6121 vrp_value_range_pool data member directly.
6122 * tree-vrp.c (simplify_stmt_for_jump_threading): Tweak slightly
6123 to use extract_range_from_stmt method to avoid need for
6124 extract_range_from_assignment method.
6125 (vrp_prop::vrp_finalize): Use set_lattice_propagation_complete
6126 method rather than setting values_propgated data member directly.
6127 * vr-values.h (class vr_values): Privatize vrp_value_range_pool,
6128 and values propagated data members and extract_range_from_assignment
6129 method. Reorder private data members to conform to standards.
6130 Add new methods set_lattice_propagation_complete and
6131 allocate_value_range.
6133 2017-11-22 Eric Botcazou <ebotcazou@adacore.com>
6135 PR rtl-optimization/83030
6136 * doc/rtl.texi (Flags in an RTL Expression): Alphabetize, add entry
6137 for CROSSING_JUMP_P and mention usage of 'jump' for JUMP_INSNs.
6138 (Insns): Delete entry for REG_CROSSING_JUMP in register notes.
6139 * bb-reorder.c (update_crossing_jump_flags): Do not test whether the
6140 CROSSING_JUMP_P flag is already set before setting it.
6141 * cfgrtl.c (fixup_partition_crossing): Likewise.
6142 * reorg.c (relax_delay_slots): Do not consider a CROSSING_JUMP_P
6145 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6147 * simplify-rtx.c (simplify_binary_operation_1) <case VEC_SERIES>:
6148 Handle the case where both arguments are using gen_const_vec_series.
6150 2017-11-22 David Malcolm <dmalcolm@redhat.com>
6153 * pretty-print.c (pp_format): Move quoting implementation to
6154 pp_begin_quote and pp_end_quote. Update pp_format_decoder call
6155 to pass address of "quote" local.
6156 (pp_begin_quote): New function.
6157 (pp_end_quote): New function.
6158 * pretty-print.h (printer_fn): Convert penultimate param from bool
6160 (pp_begin_quote): New decl.
6161 (pp_end_quote): New decl.
6162 * tree-diagnostic.c (default_tree_printer): Convert penultimate
6163 param from bool to bool *.
6164 * tree-diagnostic.h (default_tree_printer): Likewise.
6166 2017-11-22 Jeff Law <law@redhat.com>
6168 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::evrp_range_analyzer)
6169 Initialize vr_values.
6170 (evrp_range_analyzer::try_find_new_range): Call methods attached to
6171 vr_values via vr_values class instance rather than delegators.
6172 (evrp_range_analyzer::record_ranges_from_phis): Likewise.
6173 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
6174 (evrp_range_analyzer::push_value_range): Likewise.
6175 (evrp_range_analyzer::pop_value_range): Likewise.
6176 * gimple-ssa-evrp-analyze.h (class evrp_range_analyzer): Remove
6177 most delegators. Those remaining are exposed as public interfaces.
6178 Make vr_values a pointer and private.
6179 (evrp_range_analyzer::~evrp_range_analyzer): Delete the attached
6181 (evrp_range_analyzer::get_vr_value): New method.
6182 * gimple-ssa-evrp.c (class evrp_folder): Use DISABLE_COPY_AND_ASSIGN.
6183 (evrp_folder::evrp_folder): New ctor to initialize vr_values.
6184 (class evrp_dom_walker): Attach evrp_folder class, initialize
6185 it in the ctor. Remove temporary delegators.
6186 (evrp_dom_walker::before_dom_children): Call methods in attached
6187 evrp_range_analyzer class via class instance pointer. Use
6188 free value_range_constant_singleton to remove need for
6189 op_with_constant_singleton_value delegator method. Do not
6190 create a vrp_prop class instance for every call! Narrow
6191 scope of a couple variables.
6192 (evrp_dom_walker::cleanup): Call methods in attached
6193 evrp_range_analyzer class via class instance pointer.
6194 * vr-values.h (class vr_values): Privatize many methods and
6197 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6199 * tree.def (REDUC_MAX_EXPR, REDUC_MIN_EXPR, REDUC_PLUS_EXPR): Delete.
6200 * cfgexpand.c (expand_debug_expr): Remove handling for them.
6201 * expr.c (expand_expr_real_2): Likewise.
6202 * fold-const.c (const_unop): Likewise.
6203 * optabs-tree.c (optab_for_tree_code): Likewise.
6204 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
6205 * tree-inline.c (estimate_operator_cost): Likewise.
6206 * tree-pretty-print.c (dump_generic_node): Likewise.
6207 (op_code_prio): Likewise.
6208 (op_symbol_code): Likewise.
6209 * internal-fn.def (DEF_INTERNAL_SIGNED_OPTAB_FN): Define.
6210 (IFN_REDUC_PLUS, IFN_REDUC_MAX, IFN_REDUC_MIN): New internal functions.
6211 * internal-fn.c (direct_internal_fn_optab): New function.
6212 (direct_internal_fn_array, direct_internal_fn_supported_p
6213 (internal_fn_expanders): Handle DEF_INTERNAL_SIGNED_OPTAB_FN.
6214 * fold-const-call.c (fold_const_reduction): New function.
6215 (fold_const_call): Handle CFN_REDUC_PLUS, CFN_REDUC_MAX and
6217 * tree-vect-loop.c: Include internal-fn.h.
6218 (reduction_code_for_scalar_code): Rename to...
6219 (reduction_fn_for_scalar_code): ...this and return an internal
6221 (vect_model_reduction_cost): Take an internal_fn rather than
6223 (vect_create_epilog_for_reduction): Likewise. Build calls rather
6225 (vectorizable_reduction): Use internal functions rather than tree
6226 codes for the reduction operation. Update calls to the functions
6228 * config/aarch64/aarch64-builtins.c (aarch64_gimple_fold_builtin):
6229 Use calls to internal functions rather than REDUC tree codes.
6230 * config/aarch64/aarch64-simd.md: Update comment accordingly.
6232 2017-11-22 Olivier Hainque <hainque@adacore.com>
6234 * config/vxworks.c (vxworks_override_options): Pick default
6235 dwarf version from macro value, VXWORKS_DWARF_VERSION_DEFAULT.
6236 * config/vxworks.h: Define VXWORKS_DWARF_VERSION_DEFAULT and
6237 DWARF_GNAT_ENCODINGS_DEFAULT.
6238 * config/vxworksae.h: Likewise.
6240 2017-11-22 Marc Glisse <marc.glisse@inria.fr>
6242 PR tree-optimization/83104
6243 * vr-values.c (simplify_stmt_using_ranges): Check integral argument,
6246 2017-11-22 Marek Polacek <polacek@redhat.com>
6247 H.J. Lu <hongjiu.lu@intel.com>
6248 Jason Merrill <jason@redhat.com>
6253 * calls.c (initialize_argument_information): Call
6254 warn_parameter_passing_abi target hook.
6255 (store_one_arg): Use 0 for empty record size. Don't push 0 size
6256 argument onto stack.
6257 (must_pass_in_stack_var_size_or_pad): Return false for empty types.
6258 * common.opt: Update -fabi-version description.
6259 * config/i386/i386.c (init_cumulative_args): Set cum->warn_empty.
6260 (ix86_gimplify_va_arg): Call arg_int_size_in_bytes instead of
6262 (ix86_is_empty_record): New function.
6263 (ix86_warn_parameter_passing_abi): New function.
6264 (TARGET_EMPTY_RECORD_P): Redefine.
6265 (TARGET_WARN_PARAMETER_PASSING_ABI): Redefine.
6266 * config/i386/i386.h (CUMULATIVE_ARGS): Add warn_empty.
6267 * doc/tm.texi: Regenerated.
6268 * doc/tm.texi.in (TARGET_EMPTY_RECORD_P,
6269 TARGET_WARN_PARAMETER_PASSING_ABI): Add.
6270 * dwarf2out.c (get_ultimate_context): Move to tree.c.
6271 * explow.c (hard_function_value): Call arg_int_size_in_bytes
6272 instead of int_size_in_bytes.
6273 * expr.c (copy_blkmode_to_reg): Likewise.
6274 * function.c (aggregate_value_p): Return 0 for empty types.
6275 (assign_parm_find_entry_rtl): Call warn_parameter_passing_abi
6277 (locate_and_pad_parm): Call arg size_in_bytes instead
6279 * lto-streamer-out.c (hash_tree): Hash TYPE_EMPTY_P and DECL_PADDING_P.
6280 * stor-layout.c (finalize_type_size): Set TYPE_EMPTY_P.
6281 * target.def (empty_record_p, warn_parameter_passing_abi): New target
6283 * targhooks.c (hook_void_CUMULATIVE_ARGS_tree): New hook.
6284 (std_gimplify_va_arg_expr): Skip empty records. Call
6285 arg_size_in_bytes instead size_in_bytes.
6286 * targhooks.h (hook_void_CUMULATIVE_ARGS_tree): Declare.
6287 * tree-core.h (tree_type_common): Add empty_flag.
6288 (tree_decl_common): Update comments.
6289 * tree-streamer-in.c (unpack_ts_decl_common_value_fields): Stream
6291 (unpack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6292 * tree-streamer-out.c (pack_ts_decl_common_value_fields): Stream
6294 (pack_ts_type_common_value_fields): Stream TYPE_EMPTY_P.
6295 * tree.c (default_is_empty_type): New function.
6296 (default_is_empty_record): New function.
6297 (arg_int_size_in_bytes): New function.
6298 (arg_size_in_bytes): New function.
6299 (get_ultimate_context): New function.
6300 * tree.h: Define TYPE_EMPTY_P, DECL_PADDING_P and
6301 TRANSLATION_UNIT_WARN_EMPTY_P.
6302 (default_is_empty_record, arg_int_size_in_bytes,
6303 arg_size_in_bytes, get_ultimate_context): Declare.
6305 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6307 * config/arm/arm.c (cmse_clear_registers): New function.
6308 (cmse_nonsecure_call_clear_caller_saved): Replace register clearing
6309 code by call to cmse_clear_registers.
6310 (cmse_nonsecure_entry_clear_before_return): Likewise.
6312 2017-11-22 Tamar Christina <tamar.christina@arm.com>
6314 * config/arm/arm_neon.h (vdot_u32, vdotq_u32)
6315 (vdot_s32, vdotq_s32): New.
6316 (vdot_lane_u32, vdotq_lane_u32): New.
6317 (vdot_lane_s32, vdotq_lane_s32): New.
6320 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6323 * wide-int.cc (wi::add_large, wi::sub_large): Fix overflow detection
6324 for unsigned values with fewer HWIs than the precision.
6325 (test_overflow): New function.
6326 (wide_int_cc_tests): Call it.
6328 2017-11-22 Richard Sandiford <richard.sandiford@linaro.org>
6330 * emit-rtl.c (init_derived_machine_modes): Make sure ptr_mode
6331 has the same mode class as Pmode.
6333 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6335 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Get rid of
6336 padding_bits_to_clear_ptr.
6337 (cmse_nonsecure_entry_clear_before_return): Likewise.
6339 2017-11-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
6341 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Use
6342 auto_sbitap instead of integer bitfield to control register needing
6345 2017-11-22 Jakub Jelinek <jakub@redhat.com>
6347 PR tree-optimization/83044
6348 * tree-vrp.c (vrp_prop::check_array_ref): If eltsize is not
6349 INTEGER_CST or is 0, clear up_bound{,_p1} and later ignore tests
6350 that need the upper bound. Subtract offset from
6351 get_addr_base_and_unit_offset only if positive and subtract it
6352 before division by eltsize rather than after it.
6355 * valtrack.c (propagate_for_debug_subst, propagate_for_debug): Reset
6356 debug insns if they would contain UNSPEC_VOLATILE or volatile asm.
6357 (dead_debug_insert_temp): Likewise, but also ignore even non-volatile
6361 * optabs.c (expand_doubleword_mult, expand_binop): Before calling
6362 expand_binop with *mul_widen_optab, make sure at least one of the
6363 operands doesn't have VOIDmode.
6366 * dwarf2out.c (mem_loc_descriptor): Handle VEC_SERIES.
6368 PR rtl-optimization/82044
6369 PR tree-optimization/82042
6370 * dse.c (record_store): Check for overflow.
6371 (check_mem_read_rtx): Properly check for overflow if width == -1, call
6372 add_wild_read instead of clear_rhs_from_active_local_stores on
6373 overflow and log it into dump_file.
6375 2017-11-22 Richard Biener <rguenther@suse.de>
6377 * gimple-iterator.c (gimple_find_edge_insert_loc): Ignore
6378 fake edges to exit when looking for a place to insert.
6379 * tree-ssa-pre.c (clear_expression_ids): Inline into callers
6381 (insert_into_preds_of_block): Commit edge insertion immediately,
6382 assert that doesn't require new BBs.
6383 (fini_pre): Release expressions.
6384 (pass_pre::execute): Shuffle things around a bit, if the fn
6385 is too large do not compute AVAIL either as this is really the
6388 2017-11-22 Richard Biener <rguenther@suse.de>
6390 PR tree-optimization/83089
6391 * tree-if-conv.c (pass_if_conversion::execute): If anything
6392 changed reset SCEV and free the number of iteration estimates.
6394 2017-11-21 Martin Sebor <msebor@redhat.com>
6396 PR tree-optimization/82945
6397 * calls.h (warn_nonstring_bound): Remove unused function.
6399 2017-11-21 Martin Sebor <msebor@redhat.com>
6401 PR tree-optimization/82945
6402 * builtins.c (expand_builtin_strlen): Call maybe_warn_nonstring_arg.
6403 * calls.h (maybe_warn_nonstring_arg): Declare new function.
6404 * calls.c (get_attr_nonstring_decl, maybe_warn_nonstring_arg): New
6406 (initialize_argument_information): Call maybe_warn_nonstring_arg.
6407 * calls.h (get_attr_nonstring_decl): Declare new function.
6408 * doc/extend.texi (attribute nonstring): Update.
6409 * gimple-fold.c (gimple_fold_builtin_strncpy): Call
6410 get_attr_nonstring_decl and handle it.
6411 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Same. Improve
6412 detection of nul-termination.
6413 (strlen_to_stridx): Change to a pointer.
6414 (handle_builtin_strlen, handle_builtin_stxncpy): Adjust.
6415 (pass_strlen::execute): Same.
6417 2017-11-21 Sergey Shalnov <Sergey.Shalnov@intel.com>
6419 * config/i386/i386-opts.h (enum prefer_vector_width): Added new enum
6420 for the new option -mprefer-vector-width=[none|128|256|512].
6421 * config/i386/i386.c (ix86_target_string): remove old style options
6422 -mprefer-avx256 and make -mprefer-avx128 as alias.
6423 (ix86_option_override_internal): Apply defaults for the
6424 -mprefer-vector-width=[128|256] option.
6425 * config/i386/i386.h (TARGET_PREFER_AVX128, TARGET_PREFER_AVX256):
6426 Implement macros to work with -mprefer-vector-width=.
6427 * config/i386/i386.opt: Implemented option
6428 -mprefer-vector-width=[none|128|256|512].
6429 * doc/invoke.texi: Documentation for
6430 -mprefer-vector-width=[none|128|256|512].
6432 2017-11-21 Pat Haugen <pthaugen@us.ibm.com>
6434 * config/rs6000/ppc-asm.h (f50, vs50): Fix values.
6436 2017-11-21 Jonathan Wakely <jwakely@redhat.com>
6438 * doc/invoke.texi (-Wmaybe-uninitialized): Rephrase for clarity.
6440 2017-11-21 Marc Glisse <marc.glisse@inria.fr>
6442 * doc/generic.texi: Document POINTER_DIFF_EXPR, update
6444 * cfgexpand.c (expand_debug_expr): Handle POINTER_DIFF_EXPR.
6445 * expr.c (expand_expr_real_2): Likewise.
6446 * fold-const.c (const_binop, fold_addr_of_array_ref_difference,
6447 fold_binary_loc): Likewise.
6448 * match.pd (X-X, P+(Q-P), &D-P, (P+N)-P, P-(P+N), (P+M)-(P+N),
6449 P-Q==0, -(A-B), X-Z<Y-Z, (X-Z)-(Y-Z), Z-X<Z-Y, (Z-X)-(Z-Y),
6450 (A-B)+(C-A)): New transformations for POINTER_DIFF_EXPR, based on
6451 MINUS_EXPR transformations.
6452 * optabs-tree.c (optab_for_tree_code): Handle POINTER_DIFF_EXPR.
6453 * tree-cfg.c (verify_expr, verify_gimple_assign_binary): Likewise.
6454 * tree-inline.c (estimate_operator_cost): Likewise.
6455 * tree-pretty-print.c (dump_generic_node, op_code_prio,
6456 op_symbol_code): Likewise.
6457 * tree-vect-stmts.c (vectorizable_operation): Likewise.
6458 * vr-values.c (extract_range_from_binary_expr): Likewise.
6459 * varasm.c (initializer_constant_valid_p_1): Likewise.
6460 * tree.def: New tree code POINTER_DIFF_EXPR.
6462 2017-11-21 Uros Bizjak <ubizjak@gmail.com>
6464 * config/i386/i386.md (*bswap<mode>2_movbe): Add
6465 integer suffix to movbe mnemonic.
6466 (*bswaphi2_movbe): Ditto.
6467 (bswaphi_lowpart): Merge with *bswaphi_lowpart_1.
6469 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6472 * tree-cfg.c (pass_warn_function_return::execute): Formatting fix.
6473 Also warn if seen __builtin_unreachable () call with BUILTINS_LOCATION.
6474 Use LOCATION_LOCUS when comparing against UNKNOWN_LOCATION.
6476 2017-11-21 Martin Liska <mliska@suse.cz>
6478 * tree-inline.c (expand_call_inline): Remove not needed
6481 2017-11-21 James Cowgill <James.Cowgill@imgtec.com>
6482 Jakub Jelinek <jakub@redhat.com>
6485 * config/mips/frame-header-opt.c (mips_register_frame_header_opt):
6486 Remove static keyword from f variable.
6488 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6490 PR tree-optimization/83086
6491 * gimple-ssa-store-merging.c
6492 (imm_store_chain_info::try_coalesce_bswap): Test this_n.base_addr
6493 rather than n.base_addr.
6495 2017-11-21 Martin Liska <mliska@suse.cz>
6497 PR rtl-optimization/82044
6498 PR tree-optimization/82042
6499 * dse.c (check_mem_read_rtx): Check for overflow.
6501 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6503 * tree-ssa-threadbackward.c (find_jump_threads_backwards): Fix
6506 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6508 * cfgrtl.c (force_nonfallthru_and_redirect): Don't split a call
6509 and its corresponding call arg location note.
6511 2017-11-21 Claudiu Zissulescu <claziss@synopsys.com>
6512 Andrew Burgess <andrew.burgess@embecosm.com>
6514 * config/arc/arc-protos.h (arc_compute_frame_size): Delete
6516 (arc_return_slot_offset): Likewise.
6517 (arc_eh_return_address_location): New declaration.
6518 * config/arc/arc.c (TARGET_BUILTIN_SETJMP_FRAME_VALUE): Define.
6519 (MUST_SAVE_REGISTER): Add exception handler case.
6520 (MUST_SAVE_RETURN_ADDR): Likewise.
6521 (arc_frame_pointer_required): Likewise.
6522 (arc_frame_pointer_needed): New function.
6523 (arc_compute_frame_size): Changed.
6524 (arc_expand_prologue): Likewise.
6525 (arc_expand_epilogue): Likewise.
6526 (arc_initial_elimination_offset): Likewise.
6527 (arc_return_slot_offset): Delete.
6528 (arc_eh_return_address_location): New function.
6529 (arc_builtin_setjmp_frame_value): Likewise.
6530 * config/arc/arc.h (EH_RETURN_DATA_REGNO): Use 2 registers.
6531 (EH_RETURN_STACKADJ_RTX): Define.
6532 (EH_RETURN_HANDLER_RTX): Likewise.
6533 * config/arc/arc.md (eh_return): Delete.
6535 2017-11-21 Aldy Hernandez <aldyh@redhat.com>
6537 * print-tree.h (debug_vec_tree): Remove prototype.
6538 * gdbinit.in (pvt): Remove macro.
6540 2017-11-21 Jakub Jelinek <jakub@redhat.com>
6542 PR tree-optimization/83047
6543 * gimple-ssa-store-merging.c
6544 (imm_store_chain_info::output_merged_store): If the loads with the
6545 same vuse are in different basic blocks, for load_gsi pick a load
6546 location that is dominated by the other loads.
6549 * config/i386/i386.c (ix86_memmodel_check): Start
6550 -Winvalid-memory-model diagnostics with lowercase letter.
6553 * dwarf2out.c (dw_loc_list): If crtl->has_bb_partition, temporarily
6554 set in_cold_section_p to the partition containing loc_list->first.
6555 When seeing loc_list->last_before_switch node, update secname and
6556 perform range_across_switch second partition handling only after that.
6559 * run-rtl-passes.c: Include debug.h.
6560 (run_rtl_passes): Call debug_hooks->assembly_start.
6561 * dwarf2out.c (dwarf2out_assembly_start): Return early if invoked
6565 * internal-fn.c (expand_mul_overflow): Use OPTAB_WIDEN instead of
6566 OPTAB_DIRECT in calls to expand_simple_binop.
6568 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6571 * Makefile.in (C_COMMON_OBJS): Add c-family/known-headers.o.
6573 2017-11-20 David Malcolm <dmalcolm@redhat.com>
6576 * spellcheck.h (best_match::blithely_get_best_candidate): New
6579 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6581 * config/i386/i386.c (parse_mtune_ctrl_str): Start diagnostics
6582 with lower case letter.
6584 2017-11-20 Uros Bizjak <ubizjak@gmail.com>
6586 * config/i386/i386.md (bswaphi2): New expander.
6587 (*bswaphi2_movbe): New insn pattern.
6588 (bswaphi -> rorhi pepehole2): New peephole pattern.
6590 2017-11-20 Jeff Law <law@redhat.com>
6592 * Makefile.in (OBJS): Add gimple-ssa-evrp-analyze.o.
6593 * gimple-ssa-evrp-analyze.c: New file pulled from gimple-ssa-evrp.c.
6594 * gimple-ssa-evrp-analyze.h: New file pulled from gimple-ssa-evrp.c.
6595 * gimple-ssa-evrp.c: Remove bits moved into new files. Include
6596 gimple-ssa-evrp-analyze.h.
6598 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Do not
6599 set BB_VISITED here.
6600 (evrp_range_analyzer::enter): Set BB_VISITED here instead.
6602 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6604 * tree-predcom.c: Add general comment on Store-Store chains.
6605 (split_data_refs_to_components): Postpone clearing eliminate_store_p
6607 (get_chain_last_ref_at): Rename into...
6608 (get_chain_last_write_at): ...this.
6609 (get_chain_last_write_before_load): New function.
6610 (add_ref_to_chain): Promote type of chain from CT_STORE_LOAD to
6611 CT_STORE_STORE when write reference is added.
6612 (determine_roots_comp): Support load ref in CT_STORE_STORE chains.
6613 (is_inv_store_elimination_chain): Update get_chain_last_write_at call.
6614 (initialize_root_vars_store_elim_1): Ditto.
6615 (initialize_root_vars_store_elim_2): Ditto. Replace rhs once default
6616 definition is created.
6617 (execute_pred_commoning_chain): Support load ref in CT_STORE_STORE
6618 chain by replacing it with dominant stored value.
6620 2017-11-20 Bin Cheng <bin.cheng@arm.com>
6622 * tree-predcom.c (add_ref_to_chain): Remove check on distance.
6624 2017-11-20 Marc Glisse <marc.glisse@inria.fr>
6626 * vr-values.c (extract_range_from_binary_expr): Use a full range
6629 2017-11-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
6631 * config/arm/arm.md (R4_REGNUM): Define constant.
6632 (nonsecure_call_internal): Remove r4 clobber.
6633 (nonsecure_call_value_internal): Likewise.
6634 * config/arm/thumb1.md (nonsecure_call_reg_thumb1_v5): Remove second
6635 clobber and resequence match_operands.
6636 (nonsecure_call_value_reg_thumb1_v5): Likewise.
6637 * config/arm/thumb2.md (nonsecure_call_reg_thumb2): Likewise.
6638 (nonsecure_call_value_reg_thumb2): Likewise.
6640 2017-11-20 Jakub Jelinek <jakub@redhat.com>
6642 PR tree-optimization/78821
6643 * gimple-ssa-store-merging.c (find_bswap_or_nop_load): Give up
6644 if base is TARGET_MEM_REF. If base is not MEM_REF, set base_addr
6645 to the address of the base rather than the base itself.
6646 (find_bswap_or_nop_1): Just use pointer comparison for vuse check.
6647 (find_bswap_or_nop_finalize): New function.
6648 (find_bswap_or_nop): Use it.
6649 (bswap_replace): Return a tree rather than bool, change first
6650 argument from gimple * to gimple_stmt_iterator, allow inserting
6651 into an empty sequence, allow ins_stmt to be NULL - then emit
6652 all stmts into gsi. Fix up MEM_REF address gimplification.
6653 (pass_optimize_bswap::execute): Adjust bswap_replace caller.
6654 (struct store_immediate_info): Add N and INS_STMT non-static
6656 (store_immediate_info::store_immediate_info): Initialize them
6657 from newly added ctor args.
6658 (merged_store_group::apply_stores): Formatting fixes. Sort by
6660 (stmts_may_clobber_ref_p): For stores call also
6661 refs_anti_dependent_p.
6662 (gather_bswap_load_refs): New function.
6663 (imm_store_chain_info::try_coalesce_bswap): New method.
6664 (imm_store_chain_info::coalesce_immediate_stores): Use it.
6665 (split_group): Handle LROTATE_EXPR and NOP_EXPR rhs_code specially.
6666 (imm_store_chain_info::output_merged_store): Fail if number of
6667 new estimated stmts is bigger or equal than old. Handle LROTATE_EXPR
6668 and NOP_EXPR rhs_code.
6669 (pass_store_merging::process_store): Compute n and ins_stmt, if
6670 ins_stmt is non-NULL and the store rhs is otherwise invalid, use
6671 LROTATE_EXPR rhs_code. Pass n and ins_stmt to store_immediate_info
6673 (pass_store_merging::execute): Calculate dominators.
6675 * tree-ssa-math-opts.c (nop_stats, bswap_stats, struct symbolic_number,
6676 BITS_PER_MARKER, MARKER_MASK, MARKER_BYTE_UNKNOWN, HEAD_MARKER, CMPNOP,
6677 CMPXCHG, do_shift_rotate, verify_symbolic_number_p,
6678 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6679 find_bswap_or_nop_1, find_bswap_or_nop, pass_data_optimize_bswap,
6680 class pass_optimize_bswap, bswap_replace,
6681 pass_optimize_bswap::execute): Moved to ...
6682 * gimple-ssa-store-merging.c: ... this file.
6683 Include optabs-tree.h.
6684 (nop_stats, bswap_stats, do_shift_rotate, verify_symbolic_number_p,
6685 init_symbolic_number, find_bswap_or_nop_load, perform_symbolic_merge,
6686 find_bswap_or_nop_1, find_bswap_or_nop, bswap_replace): Put into
6687 anonymous namespace, remove static keywords.
6688 (pass_optimize_bswap::gate): Test BITS_PER_UNIT == 8 here...
6689 (pass_optimize_bswap::execute): ... rather than here. Formatting fix.
6691 2017-11-20 Jan Hubicka <hubicka@ucw.cz>
6694 * ipa-inline.c (can_inline_edge_p): Fix typo in previous patch.
6696 2017-11-20 Aldy Hernandez <aldyh@redhat.com>
6698 * vec.h (debug_helper): New function.
6699 (DEFINE_DEBUG_VEC): New macro.
6700 * hash-set.h (debug_helper): New function.
6701 (DEFINE_DEBUG_HASH_SET): New macro.
6702 * cfg.c (debug_slim (edge)): New function.
6703 Call DEFINE_DEBUG_VEC for edges.
6704 Call DEFINE_DEBUG_HASH_SET for edges.
6705 * cfghooks.c (debug_slim (basic_block)): New function.
6706 Call DEFINE_DEBUG_VEC for basic blocks.
6707 Call DEFINE_DEBUG_HASH_SET for basic blocks.
6708 * print-tree.c (debug_slim): New function to handle trees.
6709 Call DEFINE_DEBUG_VEC for trees.
6710 Call DEFINE_DEBUG_HASH_SET for trees.
6711 (debug (vec<tree, va_gc>) &): Remove.
6712 (debug (<vec<tree, va_gc>) *): Remove.
6713 * print-rtl.c (debug_slim): New function to handle const_rtx.
6714 Call DEFINE_DEBUG_VEC for rtx_def.
6715 Call DEFINE_DEBUG_VEC for rtx_insn.
6716 Call DEFINE_DEBUG_HASH_SET for rtx_def.
6717 Call DEFINE_DEBUG_HASH_SET for rtx_insn.
6718 * sel-sched-dump.c (debug (vec<rtx_insn *> &): Remove.
6719 (debug (vec<rtx_insn *> *ptr): Remove.
6720 (debug_insn_vector): Remove.
6721 * stor-layout.c (debug_rli): Call debug() instead of debug_vec_tree.
6723 2017-11-20 Tom de Vries <tom@codesourcery.com>
6725 PR rtl-optimization/82020
6726 * simplify-rtx.c (simplify_ternary_operation): Fix comparison mode of
6727 IF_THEN_ELSE condition.
6729 2017-11-19 Jeff Law <law@redhat.com>
6731 * tree-ssa-dom.c (record_equivalences_from_phis): Fix handling
6732 of degenerates resulting from ignoring an edge.
6734 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6737 * ipa-inline.c (can_inline_edge_p): Also check that caller is optimized
6739 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6742 * profile-count.c (profile_count::to_sreal_scale): Fix return value
6743 for uninitialied counts.
6745 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6748 * tree-inline.c (estimate_num_insns): Set to 1 at least.
6750 2017-11-19 Jan Hubicka <hubicka@ucw.cz>
6753 * i386.c (ix86_builtin_vectorization_cost): Be ready for insane types.
6755 2017-11-19 Tom de Vries <tom@codesourcery.com>
6757 * config/arc/arc.h (FUNCTION_PROFILER): Remove semicolon after
6760 2017-11-19 Tom de Vries <tom@codesourcery.com>
6762 * config/phoenix.h (TARGET_OS_CPP_BUILTINS): Remove semicolon after
6765 2017-11-19 Tom de Vries <tom@codesourcery.com>
6767 * config/visium/visium.h (ASM_OUTPUT_CASE_END): Remove semicolon after
6770 2017-11-19 Tom de Vries <tom@codesourcery.com>
6772 * config/ft32/ft32.h (REGISTER_TARGET_PRAGMAS): Remove semicolon after
6774 * config/spu/spu.h (REGISTER_TARGET_PRAGMAS): Same.
6776 2017-11-19 Tom de Vries <tom@codesourcery.com>
6778 * config/mcore/mcore-elf.h (MCORE_EXPORT_NAME): Remove semicolon after
6780 * config/mcore/mcore.h (ASM_OUTPUT_ALIGNED_COMMON): After missing
6781 semicolon after MCORE_EXPORT_NAME call.
6783 2017-11-19 Tom de Vries <tom@codesourcery.com>
6786 * vmsdbgout.c (vmsdbgout_early_finish): New function.
6787 (vmsdbg_debug_hooks): Set early_finish field to vmsdbgout_early_finish.
6789 2017-11-18 Jan Hubicka <hubicka@ucw.cz>
6791 * cgraphclones.c (cgraph_edge::clone): Rename gcov_count to prof_count.
6792 (cgraph_edge::clone): Cleanup updating of profile.
6793 * ipa-cp.c (update_profiling_info): Likewise.
6794 * ipa-inline-transform.c (inline_transform): Likewise.
6795 * ipa-inline.c (inline_small_functions): Add missing space to dump.
6796 * ipa-split.c (execute_split_functions): Do not split when function
6798 * predict.c (estimate_bb_frequencies): Cleanup updating of profile.
6799 * profile-count.c (profile_count::dump): Add global0.
6800 (profile_count::to_cgraph_frequency): Do not ICE when entry is
6802 (profile_count::to_sreal_scale): Likewise.
6803 (profile_count::adjust_for_ipa_scaling): Fix typo in comment.
6804 (profile_count::combine_with_ipa_count): New function.
6805 * profile-count.h (profile_guessed_global0adjusted): New.
6806 (profile_count::adjusted_zero): New.
6807 (profile_count::global0adjusted): New.
6808 (profile_count::combine_with_ipa_count): New.
6809 * tree-inline.c (copy_edges_for_bb): Add NUM/DEN arugment;
6810 correct profile of return block of split functions.
6811 (copy_cfg_body): Remove unused profile_count.
6812 (copy_body): Likewise.
6813 (expand_call_inline): Update.
6814 (tree_function_versioning): Update.
6816 2017-11-18 Aldy Hernandez <aldyh@redhat.com>
6818 * hash-set.h (hash_set::empty): New.
6819 * tree-ssa-threadbackward.h: Delete.
6820 * tree-ssa-threadbackward.c (class thread_jumps): New.
6821 Move max_threaded_paths into class.
6822 (fsm_find_thread_path): Remove arguments that are now in class.
6823 (profitable_jump_thread_path): Rename to...
6824 (thread_jumps::profitable_jump_thread_path): ...this.
6825 (convert_and_register_jump_thread_path): Rename to...
6826 (thread_jumps::convert_and_register_current_path): ...this.
6827 (check_subpath_and_update_thread_path): Rename to...
6828 (thread_jumps::check_subpath_and_update_thread_path): ...this.
6829 (register_jump_thread_path_if_profitable): Rename to...
6830 (thread_jumps::register_jump_thread_path_if_profitable): ...this.
6831 (handle_phi): Rename to...
6832 (thread_jumps::handle_phi): ...this.
6833 (handle_assignment): Rename to...
6834 (thread_jumps::handle_assignment): ...this.
6835 (fsm_find_control_statement_thread_paths): Rename to...
6836 (thread_jumps::fsm_find_control_statement_thread_paths): ...this.
6837 (find_jump_threads_backwards): Rename to...
6838 (thread_jumps::find_jump_threads_backwards): ...this.
6839 Initialize path local data.
6840 (pass_thread_jumps::execute): Call find_jump_threads_backwards
6841 from within thread_jumps class.
6842 (pass_early_thread_jumps::execute): Same.
6844 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6846 * cgraph.c (cgraph_node::dump): Do IPA sanity checking on IPA counts.
6848 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6850 * config/aarch64/aarch64-simd.md (fnma<mode>4): Move neg operator
6851 to canonical location.
6853 2017-11-17 Steve Ellcey <sellcey@cavium.com>
6856 * config/aarch64/aarch64.c (aarch64_use_by_pieces_infrastructure_p):
6858 (TARGET_USE_BY_PIECES_INFRASTRUCTURE_P): Remove define.
6860 2017-11-17 H.J. Lu <hongjiu.lu@intel.com>
6862 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Use
6863 rcrt1.o%s/grcrt1.o%s for -static-pie.
6865 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6867 * i386.c (ix86_multiplication_cost, ix86_division_cost,
6868 ix86_shift_rotate_cost): Break out from ...
6869 (ix86_rtx_costs): ... here.
6870 (ix86_add_stmt_cost): Use rtx cost machinery to compute cost of
6873 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6875 * predict.c (determine_unlikely_bbs): Set cgraph node count to 0
6876 when entry block was promoted unlikely.
6877 (estimate_bb_frequencies): Increase frequency scale.
6878 * profile-count.h (profile_count): Export precision info.
6880 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6882 * tree-tailcall.c (eliminate_tail_call): Be more careful about not
6883 disturbin profile of entry block.
6885 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6887 * ipa-fnsummary.c (estimate_node_size_and_time): Be more tolerant for
6890 2017-11-17 Jan Hubicka <hubicka@ucw.cz>
6892 * ipa-cp.c (update_profiling_info): Handle conversion to local profile.
6893 * tree-cfg.c (execute_fixup_cfg): Do fixup same way as inliner does.
6895 2017-11-17 Jeff Law <law@redhat.com>
6897 * gimple-ssa-evrp.c (class evrp_range_analyzer): New class extracted
6898 from evrp_dom_walker class. Various methods moved into new class.
6899 (evrp_range_analyzer::evrp_range_analyzer): Constructor for new class.
6900 (evrp_range_analyzer::enter): New method.
6901 (evrp_range_analyzer::leave): New method.
6902 (evrp_dom_walker): Remove delegators no longer needed by this class.
6903 Replace vr_values data member with evrp_range_analyzer
6905 * gimple-ssa-evrp.c (evrp_dom_walker::record_ranges_from_phis): New
6906 method extracted from evrp_dom_walker::before_dom_children.
6907 (evrp_dom_walker::record_ranges_from_stmt): Likewise.
6908 (evrp_dom_walker::record_ranges_from_incoming_edge): Likewise.
6910 * gimple-ssa-evrp.c (evrp_dom_walker): Add cleanup method.
6911 Add private copy constructor and move assignment operators.
6912 Privatize methods and class data where trivially possible.
6913 (evrp_dom_walker::cleanup): New function, extracted from
6914 execute_early_vrp. Simplify access to class data.
6916 * vr-values.h (get_output_for_vrp): Prototype.
6917 * vr-values.c (get_output_for_vrp): New function extracted from
6918 vrp_visit_assignment_or_call and extract_range_from_stmt.
6919 (vrp_visit_assignment_or_call): Use get_output_for_vrp. Simplify.
6921 2017-11-17 Luis Machado <luis.machado@linaro.org>
6923 * config/aarch64/aarch64.c
6924 (qdf24xx_prefetch_tune) <default_opt_level>: Set to -1.
6925 (qdf24xx_tunings) <autoprefetcher_model>: Set to
6926 tune_params::AUTOPREFETCHER_WEAK.
6928 2017-11-17 Tamar Christina <tamar.christina@arm.com>
6931 * config/arm/arm.c (arm_valid_target_attribute_rec):
6932 Parse "arch=" and "+<ext>".
6933 (arm_valid_target_attribute_tree): Re-init global options.
6934 (arm_option_override): Make non-static.
6935 (arm_options_perform_arch_sanity_checks): Make errors fatal.
6936 * gcc/config/arm/arm-c.c (__ARM_FEATURE_CMSE): Support undef.
6937 (__ARM_FEATURE_CRC32): Support undef.
6938 * config/arm/arm_acle.h (__ARM_FEATURE_CRC32): Replace with pragma.
6939 * doc/extend.texi (ARM Function Attributes): Add pragma and target.
6941 2017-11-17 David Malcolm <dmalcolm@redhat.com>
6943 * gdbinit.in (break-on-diagnostic): New command.
6945 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6947 * config/i386/i386.c (ix86_expand_epilogue): Change simple
6948 return to indirect jump for EH return if control-flow
6949 protection is enabled. Change explicit 'false' argument in
6950 pro_epilogue_adjust_stack with a value of
6952 * config/i386/i386.md (simple_return_indirect_internal):
6953 Remove SImode restriction to support 64-bit.
6955 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6957 * combine.c (added_notes_insn): New.
6958 (try_combine): Handle added_notes_insn like added_links_insn.
6959 Rewrite return value code.
6960 (distribute_notes): Set added_notes_insn to the earliest insn we added
6963 2017-11-17 Segher Boessenkool <segher@kernel.crashing.org>
6965 PR rtl-optimization/82621
6966 * combine.c (try_combine): Do not split PARALLELs of two SETs if the
6967 dest of one of those SETs is unused.
6969 2017-11-17 Richard Biener <rguenther@suse.de>
6972 * tree-core.h (enum annot_expr_kind): Add annot_expr_parallel_kind.
6973 * tree-pretty-print.c (dump_generic_node): Handle
6974 annot_expr_parallel_kind.
6975 * tree-cfg.c (replace_loop_annotate_in_block): Likewise.
6976 * gimplify.c (gimple_boolify): Likewise.
6978 2017-11-17 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
6980 * config.gcc (extra_headers): Add cet.h for x86 targets.
6981 * config/i386/cet.h: New file.
6982 * doc/install.texi: Add --enable-cet/--disable-cet.
6984 2017-11-17 Richard Biener <rguenther@suse.de>
6986 PR tree-optimization/83017
6987 * tree-parloops.c (MIN_PER_THREAD): Use --param parloops-min-per-thread.
6988 (gen_parallel_loop): Properly count iterations.
6989 (parallelize_loops): Handle loop->can_be_parallel independent
6990 of flag_loop_parallelize_all. Make static profitability test match
6992 * params.def (PARAM_PARLOOPS_MIN_PER_THREAD): New.
6993 * invoke.texi (parloops-min-per-thread): Document.
6995 2017-11-17 Vineet Gupta <vgupta@synopsys.com>
6997 * config/arc/linux.h: GLIBC_DYNAMIC_LINKER update per glibc
6998 upstreaming review comments.
7000 2017-11-17 Tamar Christina <tamar.christina@arm.com>
7002 * expr.c (copy_blkmode_to_reg): Fix bitsize for targets
7003 with fast unaligned access.
7004 * doc/sourcebuild.texi (word_mode_no_slow_unalign): New.
7006 2017-11-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
7008 * config/arm/arm.c (cmse_nonsecure_entry_clear_before_return): Allocate
7009 to_clear_arg_regs_bitmap to the same size as to_clear_bitmap.
7011 2017-11-17 Richard Biener <rguenther@suse.de>
7013 * tree-ssa-pre.c (phi_translate_1): Remove redundant constant
7014 folding of references.
7016 2017-11-17 Qing Zhao <qing.zhao@oracle.com>
7019 * gimple-fold.c (gimple_fold_builtin_string_compare): Add handling
7020 of replacing call to strncmp with corresponding call to strcmp when
7023 2017-11-17 Sergey Shalnov <Sergey.Shalnov@intel.com>
7025 * config/i386/x86-tune.def (X86_TUNE_AVX256_OPTIMAL): Add tuning
7026 option prefer-avx256 for skylake-avx512 configuration.
7027 * config/i386/i386.c (ix86_option_override_internal): Ditto.
7028 (get_builtin_code_for_version): Ditto.
7030 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
7031 Monk Chiang <sh.chiang04@gmail.com>
7033 * config/nds32/nds32.h (FIRST_PSEUDO_REGISTER): Modify.
7034 (FIXED_REGISTERS): Reserve more register numbers.
7035 (CALL_USED_REGISTERS): Likewise.
7036 (REG_ALLOC_ORDER): Likewise.
7037 (REG_CLASS_CONTENTS): Likewise.
7038 (REGISTER_NAMES): Likewise.
7040 2017-11-17 Chung-Ju Wu <jasonwucj@gmail.com>
7041 Kito Cheng <kito.cheng@gmail.com>
7043 * config/nds32/nds32-modes.def: Add vector mode V4QI V2HI V8QI V4HI
7045 * config/nds32/iterators.md: Add vector mode iterators and attributes.
7047 2017-11-16 Steven Munroe <munroesj@gcc.gnu.org>
7049 * config/rs6000/mmintrin.h (_mm_add_pi32[_ARCH_PWR]): Correct
7050 parameter list for vec_splats.
7052 2017-11-16 Joseph Myers <joseph@codesourcery.com>
7054 * doc/invoke.texi (-std=c17): Refer to 2018 expected publication
7056 (-std=c18, -std=iso9899:2018, -std=gnu18): Document option aliases.
7058 2017-11-16 Thomas Koenig <tkoenig@gcc.gnu.org>
7061 * doc/install.texi: Document incompatibility of Perl >=5.6.26
7062 with the required version of automake 1.11.6.
7064 2017-11-16 Pat Haugen <pthaugen@us.ibm.com>
7066 * rs6000/power9.md (power9fpdiv): New automaton and cpu_unit defined
7068 (DU_C2_3_power9): Correct reservation combinations.
7069 (FP_DIV_power9, VEC_DIV_power9): New.
7070 (power9-alu): Split out rotate/shift...
7071 (power9-rot): ...to here, correct dispatch resource.
7072 (power9-cracked-alu, power9-mul, power9-mul-compare): Correct dispatch
7074 (power9-fp): Correct latency.
7075 (power9-sdiv): Add div/sqrt resource.
7076 (power9-ddiv): Correct latency, add div/sqrt resource.
7077 (power9-sqrt, power9-dsqrt): Add div/sqrt resource.
7078 (power9-vecfdiv, power9-vecdiv): Correct latency, add div/sqrt
7080 (power9-qpdiv, power9-qpmul): Adjust resource usage.
7082 2017-11-15 Michael Meissner <meissner@linux.vnet.ibm.com>
7084 * config/rs6000/rs6000.c (rs6000_expand_builtin): Do not do the
7085 switch statement mapping KF built-ins to TF built-ins if we don't
7086 have the proper ISA 3.0 assembler support.
7088 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7090 * tree-emutls.c (lower_emutls_data): Remove unused bb_freq.
7091 (lower_emutls_function_body): Do not compute it.
7093 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7095 * ipa-split.c (split_bb_info): Turn time to sreal.
7096 (split_point): Likewise.
7097 (dump_split_point): Likewise.
7098 (fine_split_points): Likewise.
7099 (execute_split_functions): Only zero split_bbs; turn time to sreals.
7101 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7103 * ipa-fnsummary.c (analyze_function_body): Accumulate time consistently
7106 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7108 * predict.c (combine_predictions_for_bb): Preserve zero predicted
7110 (expensive_function_p): Remove useless assert.
7111 (determine_unlikely_bbs): Propagate also forward; determine cold blocks
7113 2017-11-16 Martin Sebor <msebor@redhat.com>
7115 PR tree-optimization/82588
7116 PR tree-optimization/82583
7117 * tree-vrp.c (check_array_ref): Handle flexible array members,
7118 string literals, and inner indices.
7119 (search_for_addr_array): Add detail to diagnostics.
7121 2017-11-16 Nathan Sidwell <nathan@acm.org>
7125 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
7126 (SET_DECL_ASSEMBLER_NAME): Forward to
7127 overwrite_decl_assembler_name.
7128 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
7129 (overwrite_decl_assembler_name): Declare.
7130 * tree.c (overwrite_decl_assembler_name): New.
7131 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
7132 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
7133 (LANG_HOOKS_INITIALIZER): Add it.
7134 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
7135 * langhooks.c (lhd_set_decl_assembler_name): Use
7136 SET_DECL_ASSEMBLER_NAME.
7137 (lhd_overwrite_decl_assembler_name): Default implementation.
7139 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7140 Jackson Woodruff <jackson.woodruff@arm.com>
7142 PR tree-optimization/71026
7143 * match.pd: Canonicalize constant multiplies in division.
7145 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
7147 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
7148 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
7149 Remove OPT_fomit_frame_pointer entry.
7150 * common/config/arc/arc-common.c: Likewise.
7151 * common/config/arm/arm-common.c: Likewise.
7152 * common/config/avr/avr-common.c: Likewise.
7153 * common/config/c6x/c6x-common.c: Likewise.
7154 * common/config/cr16/cr16-common.c: Likewise.
7155 * common/config/cris/cris-common.c: Likewise.
7156 * common/config/epiphany/epiphany-common.c: Likewise.
7157 * common/config/fr30/fr30-common.c: Likewise.
7158 * common/config/frv/frv-common.c: Likewise.
7159 * common/config/ia64/ia64-common.c: Likewise.
7160 * common/config/iq2000/iq2000-common.c: Likewise.
7161 * common/config/lm32/lm32-common.c: Likewise.
7162 * common/config/m32r/m32r-common.c: Likewise.
7163 * common/config/mcore/mcore-common.c: Likewise.
7164 * common/config/microblaze/microblaze-common.c: Likewise.
7165 * common/config/mips/mips-common.c: Likewise.
7166 * common/config/mmix/mmix-common.c: Likewise.
7167 * common/config/mn10300/mn10300-common.c: Likewise.
7168 * common/config/nios2/nios2-common.c: Likewise.
7169 * common/config/pa/pa-common.c: Likewise.
7170 * common/config/pdp11/pdp11-common.c: Likewise.
7171 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
7172 * common/config/riscv/riscv-common.c: Likewise.
7173 * common/config/rs6000/rs6000-common.c: Likewise.
7174 * common/config/rx/rx-common.c: Likewise.
7175 * common/config/s390/s390-common.c: Likewise.
7176 * common/config/sh/sh-common.c: Likewise.
7177 * common/config/sparc/sparc-common.c: Likewise.
7178 * common/config/tilegx/tilegx-common.c: Likewise.
7179 * common/config/tilepro/tilepro-common.c: Likewise.
7180 * common/config/v850/v850-common.c: Likewise.
7181 * common/config/visium/visium-common.c: Likewise.
7182 * common/config/xstormy16/xstormy16-common.c: Likewise.
7183 * common/config/xtensa/xtensa-common.c: Likewise.
7184 * invoke.texi (-fomit-frame-pointer): Update documentation.
7186 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
7188 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
7190 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7192 * cfg.c (scale_bbs_frequencies_int,
7193 cale_bbs_frequencies_gcov_type): Remove.
7194 * cfg.h (scale_bbs_frequencies_int,
7195 cale_bbs_frequencies_gcov_type): Remove.
7197 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7199 * tree-ssa-loop-manip.c
7200 (scale_dominated_blocks_in_loop): Update to profile counts.
7201 (tree_transform_and_unroll_loop): Likewise.
7203 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7205 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
7206 scale_bbs_frequencies_int.
7208 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7210 * final.c (compute_alignments): Use counts rather than frequencies.
7212 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7214 * cfgloopanal.c: Include sreal.h
7215 (average_num_loop_insns): Use counts and sreal for accounting.
7217 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7219 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
7222 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7223 Kito Cheng <kito.cheng@gmail.com>
7225 * config/nds32/constraints.md: Provide more constraints.
7226 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
7227 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
7228 support constraints usage.
7230 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7232 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
7234 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
7235 Kito Cheng <kito.cheng@gmail.com>
7237 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
7238 * config/nds32/nds32.opt: Refine the layout.
7239 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
7240 TARGET_EXT_STRING): Support new options.
7241 * config/nds32/nds32.h: Likewise.
7242 * config/nds32/nds32.md: Likewise.
7243 * config/nds32/nds32-predicates.c: Likewise.
7244 * config/nds32/constraints.md: Likewise.
7245 * common/config/nds32/nds32-common.c: Likewise.
7247 2017-11-16 Julia Koval <julia.koval@intel.com>
7250 * config/i386/gfniintrin.h: Add sse check.
7251 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
7253 2017-11-16 Julia Koval <julia.koval@intel.com>
7255 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
7256 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
7257 (ix86_handle_option): Handle -mavx512vbmi2.
7258 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
7259 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
7260 * config/i386/i386-c.c (__AVX512VBMI2__): New.
7261 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
7262 (ix86_valid_target_attribute_inner_p): Ditto.
7263 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
7264 * config/i386/i386.opt (mavx512vbmi2): New option.
7265 * doc/invoke.texi: Add new option.
7267 2017-11-16 Julia Koval <julia.koval@intel.com>
7269 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
7270 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
7271 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
7272 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
7273 _mm512_gf2p8mul_epi8): New intrinsics.
7274 * config/i386/i386-builtin-types.def
7275 (V64QI_FTYPE_V64QI_V64QI): New type.
7276 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
7277 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
7278 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
7279 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
7280 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
7281 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
7283 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
7285 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
7286 explicitly as a stream of bytes.
7288 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7290 * config/rs6000/altivec.h (vec_xst_be): New #define.
7291 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
7292 and externalize from *altivec_vperm_<mode>_internal.
7293 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
7295 (XL_BE_V8HI): Likewise.
7296 (XL_BE_V4SI): Likewise.
7297 (XL_BE_V4SI): Likewise.
7298 (XL_BE_V2DI): Likewise.
7299 (XL_BE_V4SF): Likewise.
7300 (XL_BE_V2DF): Likewise.
7301 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
7302 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
7303 all array entries with these keys: VSX_BUILTIN_VEC_XL,
7304 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
7305 VSX_BUILTIN_VEC_XST_BE.
7306 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
7307 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
7309 (altivec_init_builtins): Replace conditional calls to def_builtin
7310 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
7311 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
7312 with unconditional calls. Remove calls to def_builtin for
7313 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
7314 __builtin_vec_xst_be.
7315 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
7316 to define_expand, and add alternate RTL generation for P8.
7317 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
7318 vsx_ld_elemrev_v8hi.
7319 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
7320 add alternate RTL generation for P8.
7321 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
7322 vsx_ld_elemrev_v16qi.
7323 (vsx_st_elemrev_v8hi): Convert define_insn
7324 to define_expand, and add alternate RTL generation for P8.
7325 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
7326 vsx_st_elemrev_v8hi.
7327 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
7328 add alternate RTL generation for P8.
7329 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
7330 vsx_st_elemrev_v16qi.
7332 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
7335 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
7336 TARGET_AVX512ER check.
7337 (ix86_option_override_internal): Set MASK_VZEROUPPER if
7338 neither -mzeroupper nor -mno-zeroupper is used and
7339 TARGET_EMIT_VZEROUPPER is set.
7340 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
7341 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
7343 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
7345 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
7346 folding of vector compares.
7347 (fold_build_vec_cmp): New helper function.
7348 (fold_compare_helper): New helper function.
7349 (builtin_function_type): Add compare builtins to the list of functions
7350 having unsigned arguments. Cosmetic updates to comment indentation.
7351 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
7352 the not+eq combination.
7354 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7356 PR tree-optimization/82726
7357 PR tree-optimization/70754
7358 * tree-predcom.c (order_drefs_by_pos): New function.
7359 (combine_chains): Move code setting has_max_use_after to...
7360 (try_combine_chains): ...here. New parameter. Sort combined chains
7361 according to position information.
7362 (tree_predictive_commoning_loop): Update call to above function.
7363 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
7365 2017-11-15 Bin Cheng <bin.cheng@arm.com>
7367 PR tree-optimization/82726
7369 2017-01-23 Bin Cheng <bin.cheng@arm.com>
7371 PR tree-optimization/70754
7372 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
7373 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
7374 combined stmt before it if not NULL.
7375 (combine_chains): Process refs reversely and compute dominance point
7379 2017-02-23 Bin Cheng <bin.cheng@arm.com>
7381 PR tree-optimization/79663
7382 * tree-predcom.c (combine_chains): Process refs in reverse order
7383 only for ZERO length chains, and add explaining comment.
7385 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7387 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
7388 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
7389 * doc/invoke.texi (ARM Options): Add armv8.3-a.
7391 2017-11-15 Tamar Christina <tamar.christina@arm.com>
7393 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
7395 2017-11-15 Martin Liska <mliska@suse.cz>
7397 * tree-cfg.c (pass_warn_function_return::execute):
7398 Compare warn_return_type for greater than zero.
7400 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
7404 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
7405 to return true on Xeon and not on Xeon Phi.
7406 (ix86_check_avx256_register): Changed to ...
7407 (ix86_check_avx_upper_register): ... this. Add extra check for
7408 VALID_AVX512F_REG_OR_XI_MODE.
7409 (ix86_avx_u128_mode_needed): Changed
7410 ix86_check_avx256_register to ix86_check_avx_upper_register.
7411 (ix86_check_avx256_stores): Changed to ...
7412 (ix86_check_avx_upper_stores): ... this. Changed
7413 ix86_check_avx256_register to ix86_check_avx_upper_register.
7414 (ix86_avx_u128_mode_after): Changed
7415 avx_reg256_found to avx_upper_reg_found. Changed
7416 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
7417 (ix86_avx_u128_mode_entry): Changed
7418 ix86_check_avx256_register to ix86_check_avx_upper_register.
7419 (ix86_avx_u128_mode_exit): Ditto.
7420 * config/i386/i386.h: (host_detect_local_cpu): New define.
7422 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
7424 * config/arm/xgene1.md (xgene1): Split into automatons
7425 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
7426 (xgene1_f_load): Adjust reservations and/or types.
7427 (xgene1_f_store): Likewise.
7428 (xgene1_load_pair): Likewise.
7429 (xgene1_store_pair): Likewise.
7430 (xgene1_fp_load1): Likewise.
7431 (xgene1_load1): Likewise.
7432 (xgene1_store1): Likewise.
7433 (xgene1_move): Likewise.
7434 (xgene1_alu): Likewise.
7435 (xgene1_simd): Likewise.
7436 (xgene1_bfm): Likewise.
7437 (xgene1_neon_load1): Likewise.
7438 (xgene1_neon_store1): Likewise.
7439 (xgene1_neon_logic): Likewise.
7440 (xgene1_neon_st1): Likewise.
7441 (xgene1_neon_ld1r): Likewise.
7442 (xgene1_alu_cond): Added.
7443 (xgene1_shift_reg): Likwise.
7444 (xgene1_bfx): Likewise.
7445 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
7447 2017-11-15 Jakub Jelinek <jakub@redhat.com>
7450 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
7452 (can_widen_mult_without_libcall): New function.
7453 (expand_mul_overflow): If only checking unsigned mul overflow,
7454 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
7455 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
7456 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
7457 (expand_DIVMOD): Formatting fix.
7458 * expmed.h (expand_mult): Add NO_LIBCALL argument.
7459 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
7460 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
7462 PR tree-optimization/82977
7463 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
7464 constructed temporary to strlen_to_stridx.put.
7466 2017-11-15 Martin Liska <mliska@suse.cz>
7468 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
7469 * configure: Regenerate.
7471 2017-11-15 Martin Liska <mliska@suse.cz>
7474 * config/sh/sh-mem.cc: Use proper probability for
7477 2017-11-14 Jeff Law <law@redhat.com>
7479 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
7480 the red zone for stack_clash_protection_final_dynamic_probe targets
7481 when the total dynamic stack size is zero bytes.
7483 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
7484 blocks is post order.
7486 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
7488 * dumpfile.h (TDF_COMPARE_DEBUG): New.
7489 * final.c (rest_of_clean_state): Set it for the
7490 -fcompare-debug dump.
7491 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
7492 class when TDF_COMPARE_DEBUG is set.
7494 * dwarf2out.c (gen_producer_string): Discard
7497 2017-11-15 Joseph Myers <joseph@codesourcery.com>
7500 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
7501 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
7502 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
7503 (__TGMATH_REAL_2_3): Remove macros.
7504 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
7505 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
7507 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
7509 (remquo): Define using __TGMATH_REAL_3.
7511 2017-11-14 Jeff Law <law@redhat.com>
7513 * vr-values.c: New file with contents extracted from tree-vrp.c.
7514 * Makefile.in (OBJS): Add vr-values.o
7515 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
7516 (set_value_range, set_and_canonicalize_value_range): Likewise.
7517 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
7518 (value_range_constant_singleton, symbolic_range_p): Likewise.
7519 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
7520 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
7521 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
7522 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
7523 (find_case_label_range, find_case_label_index): Likewise.
7524 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
7525 (range_int_cst_singleton_p, value_inside_range): Likewise.
7526 (get_single_symbol): Likewise.
7527 (switch_update): Move structure definition here.
7528 (to_remove_edges, to_update_switch_stmts): Provide externs.
7529 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
7530 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
7531 (vrp_val_is_min, set_value_range): Likewise.
7532 (set_and_canonicalize_value_range, copy_value_range): Likewise.
7533 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
7534 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
7535 (range_is_nonnull, range_int_cst_p): Likewwise.
7536 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
7537 (get_single_symbol, operand_less_p): Likewise
7538 (compare_values_warnv, compare_values): Likewise.
7539 (value_inside_range, value_range_constant_singleton): Likewise.
7540 (zero_nonzero_bitgs_from_vr): Likewise.
7541 (extract_range_from_binary_expr_1): Likewise.
7542 (overflow_comparison_p): Likewise.
7543 (to_remove_edges, to_update_switch_stmts): Likewise.
7544 (find_case_label-index, find_case_label_range): Likewise.
7545 (switch_update, set_value_range_to_nonnegative): Remove.
7546 (set_value_range_to_truthvalue): Likewise.
7547 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
7548 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
7549 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
7550 (find_case_label_ranges, test_for_singularity): Likewise.
7551 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
7552 (x_vr_values): Move to its remaining use site.
7554 2017-11-10 Jeff Law <law@redhat.com>
7556 * vr-values.h (VR_INITIALIZER): Move #define here.
7557 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
7558 * Makefile.in (OBJS): Add tree-evrp.o
7559 * tree-vrp.h (assert_info): Move structure definition here.
7560 (set_value_range_to_varying): Prototype.
7561 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7562 (infer_value_range, register_edge_assert_for): Likewise.
7563 (stmt_interesting_for_vrp): Likewise.
7564 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
7565 (set_value_range_to_varying): No longer static.
7566 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
7567 (infer_value_range, register_edge_assert_for): Likewise.
7569 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7571 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
7572 generate the XXBRD instruction.
7574 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
7575 (rs6000_builtin_type_compatible): Treat _Float128 and long double
7576 as being compatible if -mabi=ieeelongdouble.
7577 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
7578 to setup float128 built-ins with hardware support.
7579 (BU_FLOAT128_HW_2): Likewise.
7580 (BU_FLOAT128_HW_3): Likewise.
7581 (BU_FLOAT128_HW_VSX_1): Likewise.
7582 (BU_FLOAT128_HW_VSX_2): Likewise.
7583 (scalar_extract_expq): Change float128 built-in functions to
7584 accommodate having both KFmode and TFmode functions. Use the
7585 KFmode variant as the default.
7586 (scalar_extract_sigq): Likewise.
7587 (scalar_test_neg_qp): Likewise.
7588 (scalar_insert_exp_q): Likewise.
7589 (scalar_insert_exp_qp): Likewise.
7590 (scalar_test_data_class_qp): Likewise.
7591 (sqrtf128_round_to_odd): Delete processing the round to odd
7592 built-in functions as special built-in functions, and define them
7593 as float128 built-ins. Use the KFmode variant as the default.
7594 (truncf128_round_to_odd): Likewise.
7595 (addf128_round_to_odd): Likewise.
7596 (subf128_round_to_odd): Likewise.
7597 (mulf128_round_to_odd): Likewise.
7598 (divf128_round_to_odd): Likewise.
7599 (fmaf128_round_to_odd): Likewise.
7600 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7601 support for KFmode and TFmode xststdcqp calls.
7602 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
7603 point, switch the built-in handlers for the get/set float128
7604 exponent, get float128 mantissa, float128 test built-ins, and the
7605 float128 round to odd built-in functions. Eliminate creating the
7606 float128 round to odd built-in functions as special built-ins.
7607 (rs6000_init_builtins): Eliminate special creation of the float128
7608 round to odd built-in functions.
7609 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
7610 function insns to support both TFmode and KFmode variants.
7611 (xsxsigqp_<mode>): Likewise.
7612 (xsiexpqpf_<mode>): Likewise.
7613 (xsiexpqp_<mode>): Likewise.
7614 (xststdcqp_<mode>): Likewise.
7615 (xststdcnegqp_<mode>): Likewise.
7616 (xststdcqp_<mode>): Likewise.
7618 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7620 * tree-ssa-threadupdate.c (compute_path_counts): Remove
7621 unused path_in_freq_ptr parameter.
7622 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
7624 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
7626 * ipa-inline.c (edge_badness): Dump sreal frequency.
7627 (compute_inlined_call_time): Match natural implementaiton ...
7628 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
7629 forgotten division by CGRAPH_FREQ_BASE.
7631 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7633 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
7634 Solaris 11. Update comment.
7635 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
7637 * config/sol2.h (STARTFILE_SPEC): Likewise.
7638 * configure: Regenerate.
7640 2017-11-14 Carl Love <cel@us.ibm.com>
7642 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
7643 le_ and be_ prefixes to swap* variables. Remove
7644 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
7647 2017-11-14 Jason Merrill <jason@redhat.com>
7649 Support GTY((cache)) on hash_map.
7650 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
7651 (ggc_cache_remove): Override it instead of ggc_mx.
7652 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
7653 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
7654 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
7655 (simple_cache_map_traits): Override maybe_mx.
7656 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
7657 (hash_map): Friend gt_cleare_cache.
7658 (gt_cleare_cache): New.
7659 * tree.h (tree_cache_traits): New hash_map traits class.
7660 (tree_cache_map): New typedef.
7662 2017-11-14 Richard Biener <rguenther@suse.de>
7664 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
7665 paramter and handling.
7666 (cleanup_control_flow_bb): Likewise.
7667 (cleanup_control_flow_pre): New helper performing a DFS walk
7668 to call cleanup_control_flow_bb in PRE order.
7669 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
7670 via cleanup_control_flow_pre.
7672 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
7674 * config/aarch64/aarch64-simd.md
7675 (aarch64_simd_bsl<mode>_internal): Remove DImode.
7676 (*aarch64_simd_bsl<mode>_alt): Likewise.
7677 (aarch64_simd_bsldi_internal): New.
7678 (aarch64_simd_bsldi_alt): Likewise.
7680 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7682 * tracer.c (better_p): Do not compare frequencies.
7683 * reg-stack.c (better_edge): Likewise.
7684 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
7687 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7689 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
7690 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
7691 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
7692 * lto-streamer-in.c (input_function): Use update_max_bb_count.
7693 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
7694 * predict.c (maybe_hot_frequency_p): Inline to ...
7695 (maybe_hot_count_p): ... here; rewrite to counts.
7696 (counts_to_freqs): Rename to ...
7697 (update_max_bb_count): ... this one.
7698 (expensive_function_p): Use counts.
7699 (estimate_bb_frequencies): Update.
7700 (rebuild_frequencies): Update.
7701 * predict.h (counts_to_freqs): Rename to ...
7702 (update_max_bb_count): ... this one.
7703 * profile.c (compute_branch_probabilities): Add debug info
7704 * tree-inline.c (expand_call_inline): Update debug info.
7705 (optimize_inline_calls): Use update_max_bb_count..
7706 (tree_function_versioning): Use update_max_bb_count..
7707 * value-prof.c (gimple_value_profile_transformations):
7708 Do not use update_max_bb_count.
7710 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7712 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
7713 always use frequencies.
7715 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7717 * bb-reorder.c: Remove frequencies from comments.
7718 (better_edge_p): Use profile counts.
7719 (find_traces): Dump profile counts.
7720 (rotate_loop): Use profile counts.
7721 (find_traces_1_round): Likewise.
7722 (connect_better_edge_p): Use counts instead of probabilities for
7724 (copy_bb_p): Drop early check for non-0 frequency.
7725 (sanitize_hot_paths): Update comments.
7727 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7729 * ipa-split.c (struct split_point): Add count.
7730 (consider_split): Do not compute incoming frequency; compute incoming
7731 count and store it to split_point.
7732 (split_function): Set count of the call to split part correctly.
7734 2017-11-13 Carl Love <cel@us.ibm.com>
7736 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
7738 2017-11-13 Tom Tromey <tom@tromey.com>
7740 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
7742 2017-11-13 Carl Love <cel@us.ibm.com>
7744 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
7745 Add support for builtins:
7746 unsigned int vec_first_{,miss}_match_{,or_eos}index,
7747 vector {un,}signed {char,int,short},
7748 vector {un,}signed {char,int,short}) arguments.
7749 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
7750 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
7751 Add BU_P9V_AV_2 expansions for the builtins.
7752 * config/rs6000/altivec.h (vec_first_match_index,
7753 vec_first_mismatch_index, vec_first_match_or_eos_index,
7754 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
7755 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
7756 new extern declaration.
7757 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
7758 * config/rs6000/vsx.md (first_match_index_<mode>,
7759 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
7760 first_mismatch_or_eos_index_<mode>): Add define expand.
7761 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
7762 * doc/extend.texi: Update the built-in documenation file for the new
7765 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
7767 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
7768 into the min/max operations for _Float<N> and _Float<N>X types.
7770 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7773 * dwarf2out.c (do_eh_frame): New static variable.
7774 (dwarf2out_begin_prologue): Set it.
7775 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
7777 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7779 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
7781 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
7783 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
7786 * debug.h (dwarf2out_do_eh_frame): Declare.
7787 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
7788 (dwarf2out_do_frame): Use it.
7789 (dwarf2out_do_cfi_asm): Likewise.
7790 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
7791 (dwarf2out_assembly_start): Likewise.
7792 (dwarf2out_begin_prologue): Fix comment.
7793 * toplev.c (compile_file): Always call dwarf2out_frame_finish
7794 if the target needs either debug or unwind DWARF2 info.
7795 * lto-opts.c (lto_write_options): Do not save -fexceptions,
7796 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
7797 -ftrapping-math, -ftrapv and -fwrapv.
7799 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
7801 * cgraph.c (cgraph_edge::sreal_frequency): New function.
7802 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
7803 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
7804 (estimate_edge_size_and_time): Likewise.
7805 (ipa_merge_fn_summary_after_inlining): Likewise.
7806 * ipa-inline.c (cgraph_freq_base_rec): Remove.
7807 (compute_uninlined_call_time): Use sreal_frequency.
7808 (compute_inlined_call_time): Likewise.
7809 (ipa_inline): Do not initialize cgraph_freq_base_rec.
7810 * profile-count.c: Include sreal.h.
7811 (profile_count::to_sreal_scale): New.
7812 * profile-count.h: Forward declare sreal.
7813 (profile_count::to_sreal_scale): Declare.
7815 2017-11-13 Nathan Sidwell <nathan@acm.org>
7817 * diagnostic.c (maybe_line_and_column): New.
7818 (diagnostic_get_location_text): Use it.
7819 (diagnostic_report_current_module): Likewise.
7820 (test_diagnostic_get_location_text): Add tests.
7822 2017-11-13 Luis Machado <luis.machado@linaro.org>
7824 * doc/md.texi (Specifying processor pipeline description): Fix
7825 incorrect latency for the div instruction example.
7827 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7829 PR tree-optimization/78821
7830 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
7831 that bit_not_p is the same.
7832 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
7833 (split_group): Count precisely bit_not_p bits in each statement.
7834 (invert_op): New function.
7835 (imm_store_chain_info::output_merged_store): Use invert_op to
7836 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
7837 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
7839 2017-11-13 Martin Liska <mliska@suse.cz>
7841 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
7842 (struct source_info): Likewise.
7843 (add_branch_counts): Likewise.
7844 (add_line_counts): Likewise.
7845 (function_summary): Likewise.
7846 (output_intermediate_line): Likewise.
7847 (generate_results): Likewise.
7849 2017-11-13 Martin Liska <mliska@suse.cz>
7851 * gcov.c (struct block_info): Remove typedef for block_t.
7852 (struct line_info): Likewise.
7853 (line_info::has_block): Likewise.
7854 (EXIT_BLOCK): Likewise.
7855 (unblock): Likewise.
7856 (circuit): Likewise.
7857 (get_cycles_count): Likewise.
7858 (process_file): Likewise.
7859 (read_graph_file): Likewise.
7860 (solve_flow_graph): Likewise.
7861 (find_exception_blocks): Likewise.
7862 (add_line_counts): Likewise.
7863 (accumulate_line_info): Likewise.
7864 (output_line_details): Likewise.
7866 2017-11-13 Martin Liska <mliska@suse.cz>
7868 * gcov.c (struct arc_info): Remove typedef for arc_t.
7869 (struct line_info): Likewise.
7870 (add_branch_counts): Likewise.
7871 (output_branch_count): Likewise.
7872 (function_info::~function_info): Likewise.
7873 (circuit): Likewise.
7874 (output_intermediate_line): Likewise.
7875 (read_graph_file): Likewise.
7876 (solve_flow_graph): Likewise.
7877 (find_exception_blocks): Likewise.
7878 (add_line_counts): Likewise.
7879 (accumulate_line_info): Likewise.
7880 (output_line_details): Likewise.
7881 (output_function_details): Likewise.
7883 2017-11-13 Martin Liska <mliska@suse.cz>
7885 * gcov.c (struct function_info): Remove typedef for function_t.
7886 (struct source_info): Likewise.
7887 (source_info::get_functions_at_location): Likewise.
7888 (solve_flow_graph): Likewise.
7889 (find_exception_blocks): Likewise.
7890 (add_line_counts): Likewise.
7891 (output_intermediate_file): Likewise.
7892 (process_file): Likewise.
7893 (generate_results): Likewise.
7894 (release_structures): Likewise.
7895 (read_graph_file): Likewise.
7896 (read_count_file): Likewise.
7897 (accumulate_line_counts): Likewise.
7898 (output_lines): Likewise.
7900 2017-11-13 Martin Liska <mliska@suse.cz>
7902 * gcov.c (function_info::function_info): Remove num_counts
7903 and add vector<gcov_type>.
7904 (function_info::~function_info): Use the vector.
7905 (process_file): Likewise.
7906 (read_graph_file): Likewise.
7907 (read_count_file): Likewise.
7908 (solve_flow_graph): Likewise.
7910 2017-11-13 Martin Liska <mliska@suse.cz>
7912 * gcov.c (function_info::is_artificial): New function.
7913 (process_file): Erase all artificial early.
7914 (generate_results): Skip as all artificial are already
7917 2017-11-13 Martin Liska <mliska@suse.cz>
7919 * gcov.c (read_graph_file): Store to global vector of functions.
7920 (read_count_file): Iterate the vector.
7921 (process_file): Likewise.
7922 (generate_results): Likewise.
7923 (release_structures): Likewise.
7925 2017-11-13 Jakub Jelinek <jakub@redhat.com>
7927 PR tree-optimization/82954
7928 * gimple-ssa-store-merging.c
7929 (imm_store_chain_info::coalesce_immediate_stores): If
7930 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
7932 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
7934 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
7935 Upddate call to ENDIAN_LANE_N.
7936 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
7937 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
7938 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
7939 and use aarch64_endian_lane_rtx.
7940 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
7942 2017-11-12 Tom de Vries <tom@codesourcery.com>
7944 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
7946 2017-11-12 Tom de Vries <tom@codesourcery.com>
7948 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
7950 2017-11-12 Tom de Vries <tom@codesourcery.com>
7952 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
7954 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7955 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
7956 * defaults.h (ASM_OUTPUT_LABELREF): Same.
7958 2017-11-11 Martin Sebor <msebor@redhat.com>
7961 * doc/extend.texi (attribute nonstring): Remove spurious argument.
7964 * prefic.c (translate_name): Replace strncpy with memcpy to
7965 avoid -Wstringop-truncation.
7967 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7969 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
7971 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7973 * predict.c (maybe_hot_frequency_p): Do not use cfun.
7975 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
7977 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
7980 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
7982 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
7983 enable generating XXBRH if the value is in a vector register.
7984 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
7985 value is in a vector register.
7986 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
7987 register to register bswap64's instead of doing the GPR sequence
7988 used on previous machines.
7989 (bswapdi2_xxbrd): New insn.
7990 (bswapdi2_reg): Disallow on ISA 3.0.
7991 (register to register bswap64 splitter): Do not split the insn on
7992 ISA 3.0 systems that use XXBRD.
7994 2017-11-10 Martin Sebor <msebor@redhat.com>
7997 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
7999 (find_subframework_file): Same.
8001 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
8003 * auto-profile.c (afdo_indirect_call): Drop frequency.
8004 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
8005 (cgraph_node::create_edge): Drop frequency argument.
8006 (cgraph_node::create_indirect_edge): Drop frequency argument.
8007 (cgraph_edge::make_speculative): Drop frequency arguments.
8008 (cgraph_edge::resolve_speculation): Do not update frequencies
8009 (cgraph_edge::dump_edge_flags): Do not dump frequency.
8010 (cgraph_node::dump): Check consistency in IPA mode.
8011 (cgraph_edge::maybe_hot_p): Use IPA counter.
8012 (cgraph_edge::verify_count_and_frequency): Rename to ...
8013 (cgraph_edge::verify_count): ... this one; drop frequency checking.
8014 (cgraph_node::verify_node): Update.
8015 * cgraph.h (struct cgraph_edge): Drop frequency.
8016 (cgraph_edge::frequency): New function.
8017 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
8019 (cgraph_edge::rebuild_edges): Likewise.
8020 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
8021 (duplicate_thunk_for_node): Do not pass frequency.
8022 (cgraph_node::create_clone): Scale only counts.
8023 (cgraph_node::create_virtual_clone): Do not pass frequency.
8024 (cgraph_node::create_edge_including_clones): Do not pass frequency.
8025 (cgraph_node::create_version_clone): Do not pass frequency.
8026 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
8027 (cgraph_node::expand_thunk): Do not pass frequency.
8028 (cgraph_node::create_wrapper): Do not pass frequency.
8029 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
8031 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
8032 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
8033 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
8034 (gather_caller_stats): Use frequency function.
8035 (ipcp_cloning_candidate_p): Use frequency function.
8036 (ipcp_propagate_stage): Use frequency function.
8037 (get_info_about_necessary_edges): Use frequency function.
8038 (update_profiling_info): Update only IPA profile.
8039 (update_specialized_profile): Use frequency functoin.
8040 (perhaps_add_new_callers): Update only IPA profile.
8041 * ipa-devirt.c (ipa_devirt): Use IPA profile.
8042 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
8043 (dump_ipa_call_summary): Use frequency function.
8044 (estimate_edge_size_and_time): Use frequency function.
8045 (ipa_merge_fn_summary_after_inlining): Use frequency function.
8046 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
8047 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
8048 (update_noncloned_counts): ... ths one; scale counts only.
8049 (clone_inlined_nodes): Do not scale frequency.
8050 (inline_call): Do not pass frequency.
8051 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
8052 (compute_inlined_call_time): Use IPA profile.
8053 (want_inline_small_function_p): Use IPA profile.
8054 (want_inline_self_recursive_call_p): Use IPA profile.
8055 (edge_badness): Use IPA profile.
8056 (lookup_recursive_calls): Use IPA profile.
8057 (recursive_inlining): Do not pass frequency.
8058 (resolve_noninline_speculation): Do not update frequency.
8059 (inline_small_functions): Collect max of IPA profile.
8060 (dump_overall_stats): Dump IPA porfile.
8061 (dump_inline_stats): Dump IPA porfile.
8062 (ipa_inline): Collect IPA stats.
8063 * ipa-inline.h (clone_inlined_nodes): Update prototype.
8064 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
8065 (ipa_propagate_frequency): Use frequency function.
8066 (ipa_profile): Cleanup.
8067 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
8068 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
8069 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
8070 (input_node): Do not stream frequency.
8071 (input_edge): Do not stream frequency.
8072 (merge_profile_summaries): Scale only IPA profiles.
8073 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
8074 * predict.c (drop_profile): Do not recompute frequency.
8075 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
8076 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
8077 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
8078 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
8079 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
8080 * tree-inline.c (copy_bb): Do not scale frequency.
8081 (expand_call_inline): Do not scale frequency.
8082 (tree_function_versioning): Do not scale frequency.
8083 * ubsan.c (ubsan_create_edge): Do not pass frequency.
8085 2017-11-10 Julia Koval <julia.koval@intel.com>
8087 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
8088 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
8089 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
8090 (_mm256_maskz_gf2p8affine_epi64_epi8)
8091 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
8092 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
8093 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
8094 (__builtin_ia32_vgf2p8affineqb_v32qi)
8095 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
8096 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
8098 2017-11-10 Tamar Christina <tamar.christina@arm.com>
8102 (arm_option_override): Refactor.
8103 (arm_option_reconfigure_globals): New.
8104 (arm_options_perform_arch_sanity_checks): New.
8105 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
8107 (arm_options_perform_arch_sanity_checks): Likewise
8109 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
8111 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
8112 (power9-qpmul): New.
8113 * rs6000/rs6000.md ("type" attr): Add qmul.
8114 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
8115 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
8116 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
8118 2017-11-10 Martin Sebor <msebor@redhat.com>
8121 * builtins.c (compute_objsize): Handle arrays that
8122 compute_builtin_object_size likes to fail for. Make extern.
8123 * builtins.h (compute_objsize): Declare.
8124 (check_strncpy_sizes): New function.
8125 (expand_builtin_strncpy): Call check_strncpy_sizes.
8126 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
8127 -Wstringop-truncation.
8128 (gimple_fold_builtin_strncat): Same.
8129 * gimple.c (gimple_build_call_from_tree): Set call location.
8130 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
8131 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
8132 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
8133 (handle_builtin_strlen): Use strlen_to_stridx.
8134 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
8136 Use strlen_to_stridx.
8137 (pass_strlen::execute): Release strlen_to_stridx.
8138 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
8139 (-Wstringop-truncation): Document new option.
8141 2017-11-10 Martin Liska <mliska@suse.cz>
8143 PR gcov-profile/82702
8144 * gcov.c (main): Handle intermediate files in a different
8146 (get_gcov_intermediate_filename): New function.
8147 (output_gcov_file): Remove support of intermediate files.
8148 (generate_results): Allocate intermediate file.
8149 (release_structures): Clean-up properly fn_end.
8150 (output_intermediate_file): Start iterating with line 1.
8152 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8154 PR tree-optimization/82929
8155 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8156 ops_swapped_p non-static data member.
8157 (store_immediate_info::store_immediate_info): Clear it.
8158 (imm_store_chain_info::coalesce_immediate_stores): If swapping
8159 ops set ops_swapped_p.
8160 (count_multiple_uses): Handle ops_swapped_p.
8162 2017-11-10 Martin Liska <mliska@suse.cz>
8164 * coverage.c (coverage_init): Stream information about
8165 support of has_unexecuted_blocks.
8166 * doc/gcov.texi: Document that.
8167 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
8168 * gcov.c (read_graph_file): Likewise.
8169 (output_line_beginning): Fix a small issue with
8172 2017-11-10 Bin Cheng <bin.cheng@arm.com>
8174 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
8175 reference of trivial component.
8177 2017-11-10 Jakub Jelinek <jakub@redhat.com>
8180 * gimple-ssa-store-merging.c
8181 (pass_store_merging::terminate_all_aliasing_chains): For
8182 gimple_store_p stmts also call refs_output_dependent_p.
8184 PR rtl-optimization/82913
8185 * compare-elim.c (try_merge_compare): Punt if def_insn is not
8188 2017-11-09 Jeff Law <law@redhat.com>
8190 * vr-values.h: New file with vr_values class.
8191 * tree-vrp.c: Include vr-values.h
8192 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
8193 data objects into the vr_values class.
8194 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
8195 (get_value_range): Make it a member function within vr_values class.
8196 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
8197 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
8198 (op_with_constant_singleton_value_range): Likewise.
8199 (extract_range_for_var_from_comparison_expr): Likewise.
8200 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
8201 (extract_range_from_binary_expr): Likewise.
8202 (extract_range_from_unary_expr): Likewise.
8203 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
8204 (check_for_binary_op_overflow, extract_range_basic): Likewise.
8205 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
8206 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
8207 (compare_name_with_value, compare_names): Likewise.
8208 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
8209 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
8210 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
8211 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
8212 (extract_range_from_phi_node): Likewise.
8213 (simplify_truth_ops_using_ranges): Likewise.
8214 (simplify_div_or_mod_using_ranges): Likewise.
8215 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
8216 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
8217 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
8218 (simplify_float_conversion_using_ranges): Likewise.
8219 (simplify_internal_call_using_ranges): Likewise.
8220 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
8221 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
8222 poitner via x_vr_values for calls into gimple folder.
8223 (vrp_initialize_lattice): Make this the vr_values ctor.
8224 (vrp_free_lattice): Make this the vr_values dtor.
8225 (set_vr_value): New function.
8226 (class vrp_prop): Add vr_values data member. Add various member
8227 functions as well as member functions that delegate to vr_values.
8228 (check_array_ref): Make a member function within vrp_prop class.
8229 (search_for_addr_array, vrp_initialize): Likewise.
8230 (vrp_finalize): Likewise. Revamp to avoid direct access to
8231 vr_value, values_propagated, etc.
8232 (check_array_bounds): Extract vrp_prop class instance pointer from
8233 walk info structure. Use it to call member functions.
8234 (check_all_array_refs): Make a member function within vrp_prop class.
8235 Smuggle class instance pointer via walk info structure.
8236 (x_vr_values): New local static.
8237 (vrp_valueize): Use x_vr_values to get class instance.
8238 (vr_valueize_1): Likewise.
8239 (class vrp_folder): Add vr_values data member. Add various member
8240 functions as well as member functions that delegate to vr_values.
8241 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
8242 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
8243 class instance from vr_values. Use it to call member functions.
8244 (vrp_dom_walker): Add vr_values data member.
8245 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
8246 instance via x_vr_values.
8247 (identify_jump_threads): Accept vr_values as argument. Store
8248 it into the walker structure.
8249 (evrp_dom_walker): Add vr_values class data member. Add various
8251 (evrp_dom_walker::try_find_new_range): Use vr_values data
8252 member to access the memory allocator.
8253 (evrp_dom_walker::before_dom_children): Store vr_values class
8254 instance into the vrp_folder class.
8255 (evrp_dom_walker::push_value_range): Rework to avoid direct
8256 access to num_vr_values and vr_value.
8257 (evrp_dom_walker::pop_value_range): Likewise.
8258 (execute_early_vrp): Remove call to vrp_initialize_lattice.
8259 Use vr_values to get to dump_all_value_ranges member function.
8260 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
8261 and simplify_cond_using_ranges_2 via vrp_prop class instance.
8262 Pass vr_values class instance down to identify_jump_threads.
8263 Remove call to vrp_free_lattice.
8264 (debug_all_value_ranges): Remove.
8266 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
8267 (vrp_folder): Likewise.
8269 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
8270 Get it from the existing bitmap instead.
8271 (vrp_intersect_ranges_1): Likewise.
8273 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8275 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
8277 (store_immediate_info::store_immediate_info): Add bitnotp argument,
8278 set bit_not_p to it.
8279 (imm_store_chain_info::coalesce_immediate_stores): Break group
8280 if bit_not_p is different.
8281 (count_multiple_uses, split_group,
8282 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
8283 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
8284 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
8285 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
8286 to store_immediate_info ctor.
8288 2017-11-09 Jim Wilson <jimw@sifive.com>
8290 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
8291 (scan_prog_file): Likewise.
8293 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8295 * bb-reorder.c (max_entry_frequency): Remove.
8296 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
8297 connect_traces, push_to_next_round_p): Remove prototypes.
8298 (find_traces_1_round): Use counts only.
8299 (push_to_next_round_p): Likewise.
8300 (find_traces): Likewise.
8301 (rotate_loop): Likewise.
8302 (find_traces_1_round): Likewise.
8303 (connect_traces): Likewise.
8304 (edge_order): Likewise.
8306 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
8308 * config/arm/arm.c (output_return_instruction): Add comments to
8309 indicate requirement for cmse_nonsecure_entry return to account
8310 for the size of clearing instruction output here.
8311 (thumb_exit): Likewise.
8312 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
8313 return in hardfloat mode.
8315 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
8317 * config/rs6000/rs6000.c (machine_function): Add a bool,
8318 "toc_is_wrapped_separately".
8319 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
8320 if it wasn't explicitly set or unset, we are optimizing for speed, and
8321 doing separate shrink-wrapping.
8322 (rs6000_get_separate_components): Enable the TOC component if
8323 saving the TOC register in the prologue.
8324 (rs6000_components_for_bb): Handle the TOC component.
8325 (rs6000_emit_prologue_components): Store the TOC register where needed.
8326 (rs6000_set_handled_components): Mark TOC as handled, if handled.
8327 (rs6000_emit_prologue): Don't save the TOC if that is already done.
8329 2017-11-09 Martin Jambor <mjambor@suse.cz>
8331 * ipa-param-manipulation.c: New file.
8332 * ipa-param-manipulation.h: Likewise.
8333 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
8334 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
8335 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
8336 (ipa_parm_adjustment): Likewise.
8337 (ipa_parm_adjustment_vec): Likewise.
8338 (ipa_get_vector_of_formal_parms): Moved declaration to
8339 ipa-param-manipulation.h.
8340 (ipa_get_vector_of_formal_parm_types): Likewise.
8341 (ipa_modify_formal_parameters): Likewise.
8342 (ipa_modify_call_arguments): Likewise.
8343 (ipa_combine_adjustments): Likewise.
8344 (ipa_dump_param_adjustments): Likewise.
8345 (ipa_modify_expr): Likewise.
8346 (ipa_get_adjustment_candidate): Likewise.
8347 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
8348 ipa-param-manipulation.c.
8349 (ipa_get_vector_of_formal_parm_types): Likewise.
8350 (ipa_modify_formal_parameters): Likewise.
8351 (ipa_modify_call_arguments): Likewise.
8352 (ipa_modify_expr): Likewise.
8353 (get_ssa_base_param): Likewise.
8354 (ipa_get_adjustment_candidate): Likewise.
8355 (index_in_adjustments_multiple_times_p): Likewise.
8356 (ipa_combine_adjustments): Likewise.
8357 (ipa_dump_param_adjustments): Likewise.
8358 * tree-sra.c: Also include ipa-param-manipulation.h
8359 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
8362 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8363 Alan Hayward <alan.hayward@arm.com>
8364 David Sherwood <david.sherwood@arm.com>
8366 * doc/sourcebuild.texi (vect_masked_store): Document.
8368 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8369 Alan Hayward <alan.hayward@arm.com>
8370 David Sherwood <david.sherwood@arm.com>
8372 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
8374 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8375 Alan Hayward <alan.hayward@arm.com>
8376 David Sherwood <david.sherwood@arm.com>
8378 * doc/sourcebuild.texi (vect_variable_length): Document.
8380 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8381 Alan Hayward <alan.hayward@arm.com>
8382 David Sherwood <david.sherwood@arm.com>
8384 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
8386 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8387 Alan Hayward <alan.hayward@arm.com>
8388 David Sherwood <david.sherwood@arm.com>
8390 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
8392 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8393 Alan Hayward <alan.hayward@arm.com>
8394 David Sherwood <david.sherwood@arm.com>
8396 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
8397 previously undocumented selectors.
8398 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
8400 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8402 * doc/rtl.texi (const_vector): Say that elements can be
8403 const_wide_ints too.
8404 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
8405 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
8406 (gen_vec_duplicate): Use it instead of CONSTANT_P.
8407 * optabs.c (expand_vector_broadcast): Likewise.
8409 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8410 Alan Hayward <alan.hayward@arm.com>
8411 David Sherwood <david.sherwood@arm.com>
8413 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
8414 scaled index even if the unscaled address was invalid.
8415 Don't increase the complexity of using a scale in that case.
8417 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
8418 Alan Hayward <alan.hayward@arm.com>
8419 David Sherwood <david.sherwood@arm.com>
8421 * doc/rtl.texi: Rewrite the subreg rules so that they partition
8422 the inner register into REGMODE_NATURAL_SIZE bytes rather than
8423 UNITS_PER_WORD bytes.
8424 * emit-rtl.c (validate_subreg): Divide subregs into blocks
8425 based on REGMODE_NATURAL_SIZE of the inner mode.
8426 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
8427 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
8428 * expmed.c (lowpart_bit_field_p): Divide the value up into
8429 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
8430 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
8431 whether something is likely to occupy more than one register.
8433 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
8436 * ipa-inline-transform.c (update_noncloned_frequencies): Use
8437 profile_count::adjust_for_ipa_scaling.
8438 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
8439 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
8441 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
8443 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8445 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
8446 (split_group): Add total_orig and total_new arguments, estimate the
8447 number of statements related to the store group without store merging
8448 and with store merging.
8449 (imm_store_chain_info::output_merged_store): Adjust split_group
8450 callers, punt if estimated number of statements with store merging
8451 is not smaller than estimated number of statements without it.
8453 (handled_load): Remove has_single_use checks.
8454 (pass_store_merging::process_store): Likewise.
8456 2017-11-09 Richard Biener <rguenther@suse.de>
8458 PR tree-optimization/82902
8459 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
8461 2017-11-09 Martin Liska <mliska@suse.cz>
8464 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
8467 2017-11-09 Martin Liska <mliska@suse.cz>
8469 PR tree-optimization/82669
8470 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
8472 2017-11-09 Martin Liska <mliska@suse.cz>
8474 PR gcov-profile/48463
8475 * coverage.c (coverage_begin_function): Output also end locus
8476 of a function and information whether the function is
8478 * gcov-dump.c (tag_function): Parse and print the information.
8479 * gcov.c (INCLUDE_MAP): Add include.
8480 (INCLUDE_SET): Likewise.
8481 (struct line_info): Move earlier in the source file because
8482 of vector<line_info> in function_info structure.
8483 (line_info::line_info): Likewise.
8484 (line_info::has_block): Likewise.
8485 (struct source_info): Add new member index.
8486 (source_info::get_functions_at_location): New function.
8487 (function_info::group_line_p): New function.
8488 (output_intermediate_line): New function.
8489 (output_intermediate_file): Use the mentioned function.
8490 (struct function_start): New.
8491 (struct function_start_pair_hash): Likewise.
8492 (process_file): Add code that identifies group functions.
8493 Assign lines either to global or function scope.
8494 (generate_results): Skip artificial functions.
8495 (find_source): Assign index for each source file.
8496 (read_graph_file): Read new flag artificial and end_line.
8497 (add_line_counts): Assign it either to global of function scope.
8498 (accumulate_line_counts): Isolate core of the function to
8499 accumulate_line_info and call it for both function and global
8501 (accumulate_line_info): New function.
8502 (output_line_beginning): Fix GNU coding style.
8503 (print_source_line): New function.
8504 (output_line_details): Likewise.
8505 (output_function_details): Likewise.
8506 (output_lines): Iterate both source (global) scope and function
8508 (struct function_line_start_cmp): New class.
8509 * doc/gcov.texi: Reflect changes in documentation.
8511 2017-11-09 Jakub Jelinek <jakub@redhat.com>
8514 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
8515 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
8516 and similarly for not instead of neg.
8518 2017-11-08 Andi Kleen <ak@linux.intel.com>
8520 * config/i386/i386.opt: Add -mforce-indirect-call.
8521 * config/i386/predicates.md: Check for flag_force_indirect_call.
8522 * doc/invoke.texi: Document -mforce-indirect-call
8524 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
8526 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
8528 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
8529 config/riscv/riscv.c (predict.h): New include.
8530 (riscv_slow_unaligned_access_p): No longer static.
8531 (riscv_block_move_straight): Add require.
8532 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8534 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8537 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
8538 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
8539 nonimmediate_operand predicate for operand 1 instead of
8542 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8544 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
8546 * config/aarch64/constraints.md (Uml): New constraint.
8547 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
8550 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8552 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
8553 of two vec_duplicates into a vec_concat.
8555 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8557 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8558 Simplify vec_merge of vec_duplicate and vec_concat.
8559 * config/aarch64/constraints.md (Utq): New constraint.
8560 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
8563 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8565 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
8566 Simplify vec_merge of vec_duplicate and const_vector.
8567 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
8569 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
8570 mode iterator. Update predicate on operand 1 to
8571 handle non-const_vec constants. Delete constraints.
8572 (*aarch64_combinez_be<mode>): Likewise for operand 2.
8574 2017-11-08 Jakub Jelinek <jakub@redhat.com>
8576 PR tree-optimization/78821
8577 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
8579 (store_operand_info::store_operand_info): Initialize it to false.
8580 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
8581 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
8582 store in the group, and if chain_info is non-NULL, to ignore altogether
8584 (compatible_load_p): Fail if bit_not_p does not match.
8585 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
8586 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
8587 (pass_store_merging::process_store): Adjust
8588 terminate_all_aliasing_chains calls to pass NULL in all current spots,
8589 call terminate_all_aliasing_chains newly when adding a store into
8590 a chain with non-NULL chain_info.
8592 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8594 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
8596 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
8598 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
8600 (aarch64_layout_frame): Initialise emit_frame_chain.
8601 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
8602 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
8604 2017-11-08 Martin Liska <mliska@suse.cz>
8606 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
8607 of gimple_call_internal_p.
8609 2017-11-07 Tom de Vries <tom@codesourcery.com>
8611 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
8613 2017-11-07 Tom de Vries <tom@codesourcery.com>
8615 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
8618 2017-11-08 Martin Liska <mliska@suse.cz>
8621 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
8623 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8625 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
8627 2017-11-07 Jakub Jelinek <jakub@redhat.com>
8630 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
8631 (*cmp<mode>_ccz_1): New insn with $k alternative.
8634 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
8635 RTX_COMM_COMPARE as commutative as well.
8636 (ix86_binary_operator_ok): Formatting fix.
8637 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
8638 *<code><mode>3<mask_name><round_saeonly_name>,
8639 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
8640 *<s>mul<mode>3_highpart<mask_name>,
8641 *vec_widen_umult_even_v16si<mask_name>,
8642 *vec_widen_umult_even_v8si<mask_name>,
8643 *vec_widen_umult_even_v4si<mask_name>,
8644 *vec_widen_smult_even_v16si<mask_name>,
8645 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
8646 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
8647 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
8648 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
8649 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
8650 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
8651 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
8652 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
8653 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
8654 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
8655 ix86_binary_operator_ok. Formatting fixes.
8656 (*<plusminus_insn><mode>3<mask_name><round_name>,
8657 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
8660 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
8662 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
8663 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
8664 eq and ne if TARGET_ISEL.
8665 (cmp): New code_iterator.
8666 (UNS, UNSU_, UNSIK): New code_attrs.
8667 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
8668 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
8670 ("eq<mode>3"): ... this.
8671 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
8673 ("ne<mode>3"): ... this.
8675 2017-11-07 Julia Koval <julia.koval@intel.com>
8678 * common/config/i386/i386-common.c
8679 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
8680 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
8681 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
8682 * config/i386/i386.opt: Ditto.
8683 * config/i386/i386.c (ix86_target_string): Ditto.
8684 (ix86_option_override_internal): Ditto.
8685 (ix86_init_mpx_builtins): Move MPX to args2.
8686 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
8687 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
8688 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
8689 __builtin_ia32_vgf2p8affineinvqb_v32qi,
8690 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
8691 __builtin_ia32_vgf2p8affineinvqb_v16qi,
8692 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
8694 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
8697 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
8698 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
8699 (zero-extendsidi peephole2): Remove peephole.
8701 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
8704 * stor-layout.c: Include attribs.h.
8705 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
8706 explicit lookup of "aligned" attribute.
8708 2017-11-07 Andrew Waterman <andrew@sifive.com>
8710 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
8711 (riscv_expand_block_move): Likewise.
8712 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
8714 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
8715 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
8716 * config/riscv/riscv.c (riscv_block_move_straight): New function.
8717 (riscv_adjust_block_mem): Likewise.
8718 (riscv_block_move_loop): Likewise.
8719 (riscv_expand_block_move): Likewise.
8720 * config/riscv/riscv.md (movmemsi): New pattern.
8722 2017-11-07 Michael Clark <michaeljclark@mac.com>
8724 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
8725 (MUSL_DYNAMIC_LINKER): Likewise.
8727 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
8729 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
8730 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
8733 2017-11-07 Richard Biener <rguenther@suse.de>
8735 * match.pd: Fix build.
8737 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
8738 Jackson Woodruff <jackson.woodruff@arm.com>
8740 PR tree-optimization/71026
8741 * match.pd: Canonicalize negate in division.
8743 2017-11-07 Sudakshina Das <sudi.das@arm.com>
8746 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
8748 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8750 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
8751 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
8753 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
8755 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
8756 non-scalar integral types.
8757 * match.pd (negate_expr_p): Handle MINUS_EXPR.
8758 (-(A-B), -(~A)): New transformations.
8760 2017-11-07 Tom de Vries <tom@codesourcery.com>
8762 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
8763 semicolon after "do {} while (0)".
8764 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8765 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8766 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8767 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8768 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8770 2017-11-07 Tom de Vries <tom@codesourcery.com>
8772 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
8773 after "do {} while (0)".
8774 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8775 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8776 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8777 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8778 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
8780 2017-11-07 Tom de Vries <tom@codesourcery.com>
8783 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
8785 (arm_rtx_costs_internal): Add missing semicolon after
8786 HANDLE_NARROW_SHIFT_ARITH call.
8788 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
8790 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
8791 disable isel if it was not set explicitly.
8793 2017-11-06 James Bowman <james.bowman@ftdichip.com>
8795 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE qualifiers.
8796 (add_type_attribute) likewise.
8798 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
8800 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
8801 of struct ix86_frame.
8802 (ix86_initial_elimination_offset): Likewise.
8803 (ix86_expand_split_stack_prologue): Likewise.
8805 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
8807 * tree-vrp.h (enum value_range_type): Update stale comment.
8809 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8810 Alan Hayward <alan.hayward@arm.com>
8811 David Sherwood <david.sherwood@arm.com>
8813 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
8814 (aarch64_expand_vec_perm_const): Take the number of units too.
8815 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
8816 (aarch64_expand_vec_perm_const): Likewise.
8817 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
8818 (vec_perm<mode>): Update accordingly.
8820 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8821 Alan Hayward <alan.hayward@arm.com>
8822 David Sherwood <david.sherwood@arm.com>
8824 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
8825 Take the number of units too.
8826 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
8827 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
8828 but check for a vector mode before rather than after the call.
8829 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
8830 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
8831 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
8832 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
8833 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
8834 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
8835 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
8836 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
8837 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
8838 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
8839 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
8840 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
8841 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
8842 (aarch64_sqdmull2_n<mode>): Update accordingly.
8844 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8845 Alan Hayward <alan.hayward@arm.com>
8846 David Sherwood <david.sherwood@arm.com>
8848 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
8849 the number of units too.
8850 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
8851 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
8852 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
8853 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
8854 (vec_store_lanesxi<mode>): Update accordingly.
8856 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8857 Alan Hayward <alan.hayward@arm.com>
8858 David Sherwood <david.sherwood@arm.com>
8860 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
8861 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
8862 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
8863 of units rather than the mode.
8864 * config/aarch64/iterators.md (nunits): New mode attribute.
8865 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
8866 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
8867 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
8868 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
8869 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
8870 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
8871 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
8872 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
8873 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
8874 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
8875 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
8876 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
8877 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
8878 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
8879 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
8880 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
8881 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
8882 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
8883 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
8884 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
8885 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
8886 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
8887 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
8888 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
8889 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
8890 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
8891 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
8892 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
8893 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
8894 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
8895 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
8896 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
8897 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
8898 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
8899 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
8900 (aarch64_simd_vec_setv2di): Likewise.
8902 2017-11-06 Carl Love <cel@us.ibm.com>
8904 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
8906 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
8907 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
8909 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
8911 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
8912 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
8913 Add power 8 macro expansions.
8914 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
8915 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
8916 power 8 instructions. (VSX_XXBR): Add iterator.
8918 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
8920 * config/arm/arm.md (predicable_short_it): Change default to "no",
8921 improve documentation, remove uses that are identical to the default.
8922 (enabled_for_depr_it): Rename to enabled_for_short_it.
8923 * gcc/config/arm/arm-fixed.md (predicable_short_it):
8924 Remove default uses.
8925 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
8926 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
8927 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
8928 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
8930 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
8933 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
8934 float128 helper macros, which are no longer used after deleting
8935 the old 'q' built-in functions, and moving the round to odd
8936 built-in functions to being special built-in functions.
8937 (BU_FLOAT128_2): Likewise.
8938 (BU_FLOAT128_1_HW): Likewise.
8939 (BU_FLOAT128_2_HW): Likewise.
8940 (BU_FLOAT128_3_HW): Likewise.
8941 (FABSQ): Delete old 'q' built-in functions.
8942 (COPYSIGNQ): Likewise.
8943 (SQRTF128_ODD): Move round to odd built-in functions to be
8944 special built-in functions, so that we can handle
8945 -mabi=ieeelongdouble.
8946 (TRUNCF128_ODD): Likewise.
8947 (ADDF128_ODD): Likewise.
8948 (SUBF128_ODD): Likewise.
8949 (MULF128_ODD): Likewise.
8950 (DIVF128_ODD): Likewise.
8951 (FMAF128_ODD): Likewise.
8952 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
8953 built-in names to 'f128'.
8954 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
8955 old 'q' built-in functions, as the machine independent code for
8956 'f128' built-in functions handles this.
8957 (rs6000_expand_builtin): Add expansion for float128 round to odd
8958 functions, keying off on -mabi=ieeelongdouble of whether to use
8959 the KFmode or TFmode variant.
8960 (rs6000_init_builtins): Initialize the _Float128 round to odd
8962 * doc/extend.texi (PowerPC Built-in Functions): Document the old
8963 _Float128 'q' built-in functions are now mapped into the new
8964 'f128' built-in functions.
8966 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
8968 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
8969 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
8971 2017-11-06 David Malcolm <dmalcolm@redhat.com>
8974 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
8975 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
8976 * toplev.c: Include "ipa-fnsummary.h".
8977 (toplev::finalize): Call ipa_fnsummary_c_finalize.
8979 2017-11-06 Jakub Jelinek <jakub@redhat.com>
8981 PR tree-optimization/82838
8982 * gimple-ssa-store-merging.c
8983 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
8984 on a separate gimple_seq which is then appended to seq.
8986 2017-11-06 Jeff Law <law@redhat.com>
8989 * config/i386/i386.c (PROBE_INTERVAL): Remove.
8990 (get_probe_interval): New functions.
8991 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
8992 (ix86_adjust_stack_and_probe): Likewise.
8993 (output_adjust_stack_and_probe): Likewise.
8994 (ix86_emit_probe_stack_range): Likewise.
8995 (ix86_expand_prologue): Likewise.
8997 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
8999 PR tree-optimization/82816
9000 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
9001 if the modes of the two types are the same.
9002 (convert_plusminus_to_widen): Likewise.
9004 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9006 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
9008 (usadv16qi): New define_expand.
9009 (usadv8hi): New define_expand.
9011 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
9014 * ipa-inline-transform.c (update_noncloned_frequencies): Always
9016 (inline_transform): Likewise.
9017 * predict.c (counts_to_freqs): Remove useless conditional.
9018 * profile-count.h (profile_count::apply_scale): Move sanity check.
9019 * tree-inline.c (copy_bb): Always scale.
9020 (copy_cfg_body): Likewise.
9022 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
9025 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
9027 (*cmp_ite0): Add enabled_for_depr_it attribute.
9028 (*cmp_ite1): Likewise.
9030 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
9032 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
9035 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
9037 * tree-vrp.c (vrp_int_const_binop): Return true on success and
9038 return the value by pointer.
9039 (extract_range_from_multiplicative_op_1): Update accordingly.
9040 Return as soon as an operation fails.
9042 2017-11-05 Tom de Vries <tom@codesourcery.com>
9045 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
9046 (DEF_SANITIZER_BUILTIN): ... here.
9047 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
9048 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
9050 2017-11-05 Tom de Vries <tom@codesourcery.com>
9053 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
9055 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
9056 ASM_OUTPUT_BEFORE_CASE_LABEL call.
9057 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
9059 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9060 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9061 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
9063 2017-11-05 Tom de Vries <tom@codesourcery.com>
9066 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
9069 2017-11-04 Michael Clark <michaeljclark@mac.com>
9071 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
9072 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
9074 (*addsi3_extended): Likewise.
9075 (*addsi3_extended2): Likewise.
9076 (<optab>si3): Likewise.
9077 (<optab>di3): Likewise.
9078 (<optab><mode>3): Likewise.
9079 (<*optabe>si3_internal): Likewise.
9080 (zero_extendqi<SUPERQI:mode>2): Likewise.
9081 (*add<mode>hi3): Likewise.
9082 (*xor<mode>hi3): Likewise.
9083 (<optab>di3): Likewise.
9084 (*<optab>si3_extend): Likewise.
9085 (*sge<u>_<X:mode><GPR:mode>): Likewise.
9086 (*slt<u>_<X:mode><GPR:mode>): Likewise.
9087 (*sle<u>_<X:mode><GPR:mode>): Likewise.
9089 2017-11-04 Andrew Waterman <andrew@sifive.com>
9091 * config/riscv/riscv.c (riscv_option_override): Conditionally set
9092 TARGET_STRICT_ALIGN based upon -mtune argument.
9094 2017-11-04 Andrew Waterman <andrew@sifive.com>
9096 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
9098 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
9100 * config/i386/i386.c (choose_basereg): Use optional scratch
9101 register and add assertion.
9102 (x86_emit_outlined_ms2sysv_save): Use scratch register when
9103 needed, and don't allocate stack.
9104 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
9105 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
9106 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
9108 2017-11-03 Jeff Law <law@redhat.com>
9110 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
9111 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
9112 to probe at the start of a noreturn function.
9114 2017-11-03 Jakub Jelinek <jakub@redhat.com>
9116 PR tree-optimization/78821
9117 * gimple-ssa-store-merging.c: Update the file comment.
9118 (MAX_STORE_ALIAS_CHECKS): Define.
9119 (struct store_operand_info): New type.
9120 (store_operand_info::store_operand_info): New constructor.
9121 (struct store_immediate_info): Add rhs_code and ops data members.
9122 (store_immediate_info::store_immediate_info): Add rhscode, op0r
9123 and op1r arguments to the ctor, initialize corresponding data members.
9124 (struct merged_store_group): Add load_align_base and load_align
9126 (merged_store_group::merged_store_group): Initialize them.
9127 (merged_store_group::do_merge): Update them.
9128 (merged_store_group::apply_stores): Pick the constant for
9129 encode_tree_to_bitpos from one of the two operands, or skip
9130 encode_tree_to_bitpos if neither operand is a constant.
9131 (class pass_store_merging): Add process_store method decl. Remove
9132 bool argument from terminate_all_aliasing_chains method decl.
9133 (pass_store_merging::terminate_all_aliasing_chains): Remove
9134 var_offset_p argument and corresponding handling.
9135 (stmts_may_clobber_ref_p): New function.
9136 (compatible_load_p): New function.
9137 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
9138 if there is overlap and rhs_code is not INTEGER_CST. For
9139 non-overlapping stores terminate group if rhs is not mergeable.
9140 (get_alias_type_for_stmts): Change first argument from
9141 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
9142 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
9143 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
9145 (get_location_for_stmts): Change first argument from
9146 auto_vec<gimple *> & to vec<gimple *> &.
9147 (struct split_store): Remove orig_stmts data member, add orig_stores.
9148 (split_store::split_store): Create orig_stores rather than orig_stmts.
9149 (find_constituent_stmts): Renamed to ...
9150 (find_constituent_stores): ... this. Change second argument from
9151 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
9152 to info structures rather than the statements.
9153 (split_group): Rename ALLOW_UNALIGNED argument to
9154 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
9155 it. Adjust find_constituent_stores caller.
9156 (imm_store_chain_info::output_merged_store): Handle rhs_code other
9157 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
9158 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
9159 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
9160 (mem_valid_for_store_merging): New function.
9161 (handled_load): New function.
9162 (pass_store_merging::process_store): New method.
9163 (pass_store_merging::execute): Use process_store method. Adjust
9164 terminate_all_aliasing_chains caller.
9166 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9168 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
9169 Return true for more constants, symbols and label references.
9170 (aarch64_valid_floating_const): Remove unused function.
9172 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9175 * config/aarch64/aarch64.c (aarch64_layout_frame):
9176 Undo forcing of LR at bottom of frame.
9178 2017-11-03 Jeff Law <law@redhat.com>
9181 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
9182 for int_registers_saved.
9184 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
9185 extracted from tree-ssa-dom.c.
9186 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
9187 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
9188 (record_equivalences_from_incoming_edge): Add additional argument
9189 to single_pred_edge_ignoring_loop_edges call.
9190 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
9191 (uncprop_dom_walker::before_dom_children): Add additional argument
9192 to single_pred_edge_ignoring_loop_edges call.
9193 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
9194 single_pred_edge_ignoring_loop_edges rather than open coding.
9195 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
9197 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
9199 * match.pd (-(-A)): Rewrite.
9201 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9203 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
9204 (rs6000_emit_int_cmove): New declaration.
9205 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
9206 (rs6000_emit_sISEL): Delete.
9207 (rs6000_emit_int_cmove): Make non-static.
9208 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
9209 instead of rs6000_emit_sISEL.
9211 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
9213 * asan.c (create_cond_insert_point): Maintain profile.
9214 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
9216 * basic-block.h (struct basic_block_def): Remove frequency.
9217 (EDGE_FREQUENCY): Use to_frequency
9218 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
9220 (find_traces): Update to use to_frequency.
9221 (find_traces_1_round): Likewise; use only IPA counts.
9222 (bb_to_key): Likewise.
9223 (connect_traces): Use IPA counts only.
9224 (copy_bb_p): Update to use to_frequency.
9225 (fix_up_crossing_landing_pad): Likewise.
9226 (sanitize_hot_paths): Likewise.
9227 * bt-load.c (basic_block_freq): Likewise.
9228 * cfg.c (init_flow): Set count_max to uninitialized.
9229 (check_bb_profile): Remove frequencies; check counts.
9230 (dump_bb_info): Do not dump frequencies.
9231 (update_bb_profile_for_threading): Update counts only.
9232 (scale_bbs_frequencies_int): Likewise.
9233 (MAX_SAFE_MULTIPLIER): Remove.
9234 (scale_bbs_frequencies_gcov_type): Update counts only.
9235 (scale_bbs_frequencies_profile_count): Update counts only.
9236 (scale_bbs_frequencies): Update counts only.
9237 * cfg.h (struct control_flow_graph): Add count-max.
9238 (update_bb_profile_for_threading): Update prototype.
9239 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
9240 (find_many_sub_basic_blocks): Likewise.
9241 * cfgcleanup.c (try_forward_edges): Likewise.
9242 (try_crossjump_to_edge): Likewise.
9243 * cfgexpand.c (expand_gimple_cond): Likewise.
9244 (expand_gimple_tailcall): Likewise.
9245 (construct_init_block): Likewise.
9246 (construct_exit_block): Likewise.
9247 * cfghooks.c (verify_flow_info): Check consistency of counts.
9248 (dump_bb_for_graph): Do not dump frequencies.
9249 (split_block_1): Do not update frequencies.
9250 (split_edge): Do not update frequencies.
9251 (make_forwarder_block): Do not update frequencies.
9252 (duplicate_block): Do not update frequencies.
9253 (account_profile_record): Do not update frequencies.
9254 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
9255 for global heuristics.
9256 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
9257 (expected_loop_iterations_unbounded): Use counts only.
9258 * cfgloopmanip.c (scale_loop_profile): Simplify.
9259 (create_empty_loop_on_edge): Simplify
9261 (duplicate_loop_to_header_edge): Simplify
9262 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
9263 (update_br_prob_note): Take care of removing note when profile
9265 (relink_block_chain): Do not dump frequency.
9266 (rtl_account_profile_record): Use to_frequency.
9267 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
9268 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
9269 (cgraph_update_edges_for_call_stmt_node): Likewise.
9270 (cgraph_edge::verify_count_and_frequency): Update.
9271 (cgraph_node::verify_node): Temporarily disable frequency verification.
9272 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
9273 to_cgraph_frequency.
9274 (cgraph_edge::rebuild_edges): Convert to ipa counts.
9275 * cgraphunit.c (init_lowered_empty_function): Do not initialize
9277 (cgraph_node::expand_thunk): Update profile.
9278 * except.c (dw2_build_landing_pads): Do not update frequency.
9279 * final.c (compute_alignments): Use to_frequency.
9280 (dump_basic_block_info): Do not dump frequency.
9281 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
9282 (dump_gimple_bb_header): Do not dump frequency.
9283 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
9285 * gimple-streamer-in.c (input_bb): Do not stream frequency.
9286 * gimple-streamer-out.c (output_bb): Do not stream frequency.
9287 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
9288 (init_before_recovery): Do not update frequency.
9289 (sched_create_recovery_edges): Do not update frequency.
9290 * hsa-gen.c (convert_switch_statements): Do not update frequency.
9291 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
9292 (ipa_cp_c_finalize): Set max_count to uninitialized.
9293 * ipa-fnsummary.c (get_minimal_bb): Use counts.
9294 (param_change_prob): Use counts.
9295 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
9297 * ipa-split.c (consider_split): Use to_frequency.
9298 (split_function): Use to_frequency.
9299 * ira-build.c (loop_compare_func): Likewise.
9300 (mark_loops_for_removal): Likewise.
9301 (mark_all_loops_for_removal): Likewise.
9302 * loop-doloop.c (doloop_modify): Do not update frequency.
9303 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
9305 * lto-streamer-in.c (input_function): Update count_max.
9306 * omp-expand.c (expand_omp_taskreg): Update count_max.
9307 * omp-simd-clone.c (simd_clone_adjust): Update profile.
9308 * predict.c (maybe_hot_frequency_p): Use to_frequency.
9309 (maybe_hot_count_p): Use ipa counts only.
9310 (maybe_hot_bb_p): Simplify.
9311 (maybe_hot_edge_p): Simplify.
9312 (probably_never_executed): Do not take frequency argument.
9313 (probably_never_executed_bb_p): Do not pass frequency.
9314 (probably_never_executed_edge_p): Likewise.
9315 (combine_predictions_for_bb): Check that profile is nonzero.
9316 (propagate_freq): Do not set frequency.
9317 (drop_profile): Simplify.
9318 (counts_to_freqs): Simplify.
9319 (expensive_function_p): Use to_frequency.
9320 (propagate_unlikely_bbs_forward): Simplify.
9321 (determine_unlikely_bbs): Simplify.
9322 (estimate_bb_frequencies): Add hack to silence graphite issues.
9323 (compute_function_frequency): Use ipa counts.
9324 (pass_profile::execute): Update.
9325 (rebuild_frequencies): Use counts only.
9326 (force_edge_cold): Use counts only.
9327 * profile-count.c (profile_count::dump): Dump new count types.
9328 (profile_count::differs_from_p): Check compatiblity.
9329 (profile_count::to_frequency): New function.
9330 (profile_count::to_cgraph_frequency): New function.
9331 * profile-count.h (struct function): Declare.
9332 (enum profile_quality): Add profile_guessed_local and
9333 profile_guessed_global0.
9334 (class profile_proability): Decrease number of bits to 29;
9335 update from_reg_br_prob_note and to_reg_br_prob_note.
9336 (class profile_count: Update comment; decrease number of bits
9337 to 61. Check compatibility.
9338 (profile_count::compatible_p): New private member function.
9339 (profile_count::ipa_p): New member function.
9340 (profile_count::operator<): Handle global zero correctly.
9341 (profile_count::operator>): Handle global zero correctly.
9342 (profile_count::operator<=): Handle global zero correctly.
9343 (profile_count::operator>=): Handle global zero correctly.
9344 (profile_count::nonzero_p): New member function.
9345 (profile_count::force_nonzero): New member function.
9346 (profile_count::max): New member function.
9347 (profile_count::apply_scale): Handle IPA scalling.
9348 (profile_count::guessed_local): New member function.
9349 (profile_count::global0): New member function.
9350 (profile_count::ipa): New member function.
9351 (profile_count::to_frequency): Declare.
9352 (profile_count::to_cgraph_frequency): Declare.
9353 * profile.c (OVERLAP_BASE): Delete.
9354 (compute_frequency_overlap): Delete.
9355 (compute_branch_probabilities): Do not use compute_frequency_overlap.
9356 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
9357 * sched-ebb.c (rank): Use counts only.
9358 * shrink-wrap.c (handle_simple_exit): Use counts only.
9359 (try_shrink_wrapping): Use counts only.
9360 (place_prologue_for_one_component): Use counts only.
9361 * tracer.c (find_best_predecessor): Use to_frequency.
9362 (find_trace): Use to_frequency.
9363 (tail_duplicate): Use to_frequency.
9364 * trans-mem.c (expand_transaction): Do not update frequency.
9365 * tree-call-cdce.c: Do not update frequency.
9366 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
9367 (gimple_merge_blocks): Likewise.
9368 (gimple_split_edge): Likewise.
9369 (gimple_duplicate_sese_region): Likewise.
9370 (gimple_duplicate_sese_tail): Likewise.
9371 (move_sese_region_to_fn): Likewise.
9372 (gimple_account_profile_record): Likewise.
9373 (insert_cond_bb): Likewise.
9374 * tree-complex.c (expand_complex_div_wide): Likewise.
9375 * tree-eh.c (lower_resx): Update profile.
9376 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
9378 (initialize_cfun): Do not initialize frequencies
9379 (freqs_to_counts): Delete.
9380 (copy_cfg_body): Ignore count parameter.
9381 (copy_body): Update.
9382 (expand_call_inline): Update count_max.
9383 (optimize_inline_calls): Update count_max.
9384 (tree_function_versioning): Update count_max.
9385 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
9386 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
9388 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
9389 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
9390 (try_peel_loop): Likewise.
9391 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
9393 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
9394 (tree_transform_and_unroll_loop): Do not use frequencies
9395 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
9396 Use reliable prediction only.
9397 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
9398 * tree-ssa-sink.c (select_best_block): Use to_frequency.
9399 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
9400 probability scaling.
9401 * tree-ssa-threadupdate.c (create_block_for_threading): Do
9402 not update frequency
9403 (any_remaining_duplicated_blocks): Likewise.
9404 (update_profile): Likewise.
9405 (estimated_freqs_path): Delete.
9406 (freqs_to_counts_path): Delete.
9407 (clear_counts_path): Delete.
9408 (ssa_fix_duplicate_block_edges): Likewise.
9409 (duplicate_thread_path): Likewise.
9410 * tree-switch-conversion.c (gen_inbound_check): Use counts.
9411 * tree-tailcall.c (decrease_profile): Do not update frequency.
9412 (eliminate_tail_call): Likewise.
9413 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
9414 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
9415 (optimize_mask_stores): Likewise.
9416 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
9417 * ubsan.c (ubsan_expand_null_ifn): Update profile.
9418 (ubsan_expand_ptr_ifn): Update profile.
9419 * value-prof.c (gimple_ic): Simplify.
9420 * value-prof.h (gimple_ic): Update prototype.
9421 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
9422 * ipa-inline.c (compute_uninlined_call_time): Be sure that
9424 (want_inline_self_recursive_call_p): Likewise.
9425 (resolve_noninline_speculation): Only cummulate defined counts.
9426 (inline_small_functions): Use nonzero_p.
9427 (ipa_inline): Do not access freed node.
9429 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
9431 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9432 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
9434 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
9436 * config/riscv/riscv.c (riscv_legitimize_move): Handle
9437 non-legitimate address.
9439 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9441 * config/rs6000/rs6000.md (*lt0_disi): Delete.
9442 (*lt0_<mode>di, *lt0_<mode>si): New.
9444 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
9446 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
9447 TARGET_PAIRED_FLOAT.
9449 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
9450 Jim Wilson <jim.wilson@linaro.org>
9452 * config/aarch64/aarch64-cores.def (saphira): New CPU.
9453 * config/aarch64/aarch64-tune.md: Regenerated.
9454 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
9455 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
9457 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
9459 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
9460 (arc_expand_prologue): Restore blink for millicode.
9461 * config/arc/linux.h (LINK_EH_SPEC): Defined.
9463 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
9466 * config/i386/i386.c (ix86_vector_duplicate_value): Use
9467 gen_vec_duplicate after forcing the scalar into a register.
9469 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
9471 * combine (try_combine): Print the insns input to try_combine to the
9474 2017-11-02 Steve Ellcey <sellcey@cavium.com>
9477 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
9478 Remove second argument from aarch64_process_target_attr call.
9479 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
9481 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
9483 (aarch64_handle_attr_arch): Remove second argument.
9484 (aarch64_handle_attr_cpu): Ditto.
9485 (aarch64_handle_attr_tune): Ditto.
9486 (aarch64_handle_attr_isa_flags): Ditto.
9487 (aarch64_process_one_target_attr): Ditto.
9488 (aarch64_process_target_attr): Ditto.
9489 (aarch64_option_valid_attribute_p): Remove second argument.
9490 on aarch64_process_target_attr call.
9492 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9494 * diagnostic.c: Include "selftest-diagnostic.h".
9495 (selftest::assert_location_text): New function.
9496 (selftest::test_diagnostic_get_location_text): New function.
9497 (selftest::diagnostic_c_tests): Call it.
9499 2017-11-02 David Malcolm <dmalcolm@redhat.com>
9501 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
9502 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
9503 (class selftest::test_diagnostic_context): Move to...
9504 * selftest-diagnostic.c: New file.
9505 * selftest-diagnostic.h: New file.
9507 2017-11-02 James Bowman <james.bowman@ftdichip.com>
9509 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
9510 offset range for FT32B.
9511 * config/ft32/ft32.h: option "mcompress" enables relaxation.
9512 * config/ft32/ft32.md: Add TARGET_NOPM.
9513 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
9514 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
9516 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
9518 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
9520 2017-11-02 Jeff Law <law@redhat.com>
9522 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
9523 virtual keyword on FINAL OVERRIDE members.
9525 * tree-ssa-propagate.h (ssa_propagation_engine): Group
9526 virtuals together. Add virtual destructor.
9527 (substitute_and_fold_engine): Similarly.
9529 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
9531 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
9533 2017-11-02 Richard Biener <rguenther@suse.de>
9535 PR tree-optimization/82795
9536 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
9538 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9540 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
9542 (gcc_GAS_CHECK_FEATURE): Remove.
9543 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
9544 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
9545 * configure: Regenerate.
9547 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
9549 * config/arc/arc.c (hwloop_optimize): Account for empty
9552 2017-11-02 Richard Biener <rguenther@suse.de>
9555 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
9556 Truncate ARRAY_REF index and element size.
9558 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9560 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
9562 2017-11-01 Jeff Law <law@redhat.com>
9564 * tree-ssa-ccp.c (ccp_folder): New class derived from
9565 substitute_and_fold_engine.
9566 (ccp_folder::get_value): New member function.
9567 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
9568 (ccp_fold_stmt): Remove prototype.
9569 (ccp_finalize): Call substitute_and_fold from the ccp_class.
9570 * tree-ssa-copy.c (copy_folder): New class derived from
9571 substitute_and_fold_engine.
9572 (copy_folder::get_value): Renamed from get_value.
9573 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
9574 * tree-vrp.c (vrp_folder): New class derived from
9575 substitute_and_fold_engine.
9576 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
9577 (vrp_folder::get_value): New member function.
9578 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
9579 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
9580 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
9581 provide a class interface to folder/substitute routines.
9582 (ssa_prop_fold_stmt_fn): Remove typedef.
9583 (ssa_prop_get_value_fn): Likewise.
9584 (subsitute_and_fold): Remove prototype.
9585 (replace_uses_in): Likewise.
9586 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
9587 Renamed from replace_uses_in. Call the virtual member function
9588 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
9589 (substitute_and_fold_dom_walker): Remove initialization of
9590 data member entries for calbacks. Add substitute_and_fold_engine
9591 member and initialize it.
9592 (substitute_and_fold_dom_walker::before_dom_children0: Use the
9593 member functions for get_value, replace_phi_args_in c
9594 replace_uses_in, and fold_stmt calls.
9595 (substitute_and_fold_engine::substitute_and_fold): Renamed from
9596 substitute_and_fold. Remove assert. Update ctor call.
9598 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
9599 (ssa_prop_visit_phi_fn): Likewise.
9600 (class ssa_propagation_engine): New class to provide an interface
9602 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
9604 (ssa_prop_visit_phi): Likewise.
9605 (ssa_propagation_engine::simulate_stmt): Moved into class.
9606 Call visit_phi/visit_stmt from the class rather than via
9607 file scoped static variables.
9608 (ssa_propagation_engine::simulate_block): Moved into class.
9609 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
9610 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
9611 set file scoped statics for the visit_stmt/visit_phi callbacks.
9612 * tree-complex.c (complex_propagate): New class derived from
9613 ssa_propagation_engine.
9614 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
9615 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
9616 (tree_lower_complex): Call ssa_propagate via the complex_propagate
9618 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
9619 ssa_propagation_engine.
9620 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
9621 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
9622 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
9623 * tree-ssa-copy.c (copy_prop): New class derived from
9624 ssa_propagation_engine.
9625 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
9626 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
9627 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
9628 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
9629 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
9630 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
9631 (execute_vrp): Call ssa_propagate from the vrp_prop class.
9633 2017-11-01 Jakub Jelinek <jakub@redhat.com>
9635 PR rtl-optimization/82778
9636 PR rtl-optimization/82597
9637 * compare-elim.c (struct comparison): Add in_a_setter field.
9638 (find_comparison_dom_walker::before_dom_children): Remove killed
9639 bitmap and df_simulate_find_defs call, instead walk the defs.
9640 Compute last_setter and initialize in_a_setter. Merge definitions
9641 with first initialization for a few variables.
9642 (try_validate_parallel): Use insn_invalid_p instead of
9643 recog_memoized. Return insn rather than just the pattern.
9644 (try_merge_compare): Fix up comment. Don't uselessly test if
9645 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
9647 (execute_compare_elim_after_reload): Remove df_chain_add_problem
9650 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9651 Alan Hayward <alan.hayward@arm.com>
9652 David Sherwood <david.sherwood@arm.com>
9654 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
9655 aarch64_hard_regno_nregs to get the number of registers
9658 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9659 Alan Hayward <alan.hayward@arm.com>
9660 David Sherwood <david.sherwood@arm.com>
9662 * config/aarch64/constraints.md (Upl): Rename to...
9664 * config/aarch64/aarch64.md
9665 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
9668 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9669 Alan Hayward <alan.hayward@arm.com>
9670 David Sherwood <david.sherwood@arm.com>
9672 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
9673 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
9676 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9677 Alan Hayward <alan.hayward@arm.com>
9678 David Sherwood <david.sherwood@arm.com>
9680 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
9681 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
9682 (aarch64_evpc_dup): Generate rtl direcly, rather than using
9684 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
9685 of a single element.
9686 * config/aarch64/iterators.md: Add a comment above the permute
9687 unspecs to say that they are generated directly by
9688 aarch64_expand_vec_perm_const.
9689 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
9691 2017-11-01 Nathan Sidwell <nathan@acm.org>
9693 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
9695 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
9697 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
9698 and medany code models, and describe what they do.
9700 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9702 Revert accidental duplicate:
9704 * combine.c (can_change_dest_mode): Reject changes in
9705 REGMODE_NATURAL_SIZE.
9707 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
9709 PR rtl-optimization/64682
9710 PR rtl-optimization/69567
9711 PR rtl-optimization/69737
9712 PR rtl-optimization/82683
9713 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
9714 register mentioned in the note, drop the note, unless it came from I3,
9715 in which case it should go to I3 again.
9717 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9719 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
9720 and return false if not.
9721 (clear_bytes_written_by, live_bytes_read): Update accordingly.
9723 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9725 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
9726 range is known to be empty.
9728 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9729 Alan Hayward <alan.hayward@arm.com>
9730 David Sherwood <david.sherwood@arm.com>
9732 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
9733 and CONST_VECTOR_NUNITS instead of computing the number of units from
9734 the byte sizes of the vector and element.
9735 (simplify_binary_operation_1): Likewise.
9736 (simplify_const_binary_operation): Likewise.
9737 (simplify_ternary_operation): Likewise.
9739 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9740 Alan Hayward <alan.hayward@arm.com>
9741 David Sherwood <david.sherwood@arm.com>
9743 * var-tracking.c (INT_MEM_OFFSET): Replace with...
9744 (int_mem_offset): ...this new function.
9745 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
9746 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
9747 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
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 * lower-subreg.c (interesting_mode_p): New function.
9755 (compute_costs, find_decomposable_subregs, decompose_register)
9756 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
9757 (resolve_clobber, dump_choices): Use it.
9759 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9760 Alan Hayward <alan.hayward@arm.com>
9761 David Sherwood <david.sherwood@arm.com>
9763 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
9765 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9766 Alan Hayward <alan.hayward@arm.com>
9767 David Sherwood <david.sherwood@arm.com>
9769 * alias.c (find_base_value, find_base_term): Only process integer
9770 truncations. Check the precision rather than the size.
9772 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9773 Alan Hayward <alan.hayward@arm.com>
9774 David Sherwood <david.sherwood@arm.com>
9776 * machmode.h (is_narrower_int_mode): New function
9777 * optabs.c (expand_float, expand_fix): Use it.
9778 * dwarf2out.c (rotate_loc_descriptor): Likewise.
9780 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9781 Alan Hayward <alan.hayward@arm.com>
9782 David Sherwood <david.sherwood@arm.com>
9784 * rtl.h (narrower_subreg_mode): New function.
9785 * ira-color.c (update_costs_from_allocno): Use it.
9787 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9788 Alan Hayward <alan.hayward@arm.com>
9789 David Sherwood <david.sherwood@arm.com>
9791 * optabs-query.h (convert_optab_p): New function, split out from...
9792 (convert_optab_handler): ...here.
9793 (widening_optab_handler): Delete.
9794 (find_widening_optab_handler): Remove permit_non_widening parameter.
9795 (find_widening_optab_handler_and_mode): Likewise. Provide an
9796 override that operates on mode class wrappers.
9797 * optabs-query.c (widening_optab_handler): Delete.
9798 (find_widening_optab_handler_and_mode): Remove permit_non_widening
9799 parameter. Assert that the two modes are the same class and that
9800 the "from" mode is narrower than the "to" mode. Use
9801 convert_optab_handler instead of widening_optab_handler.
9802 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
9803 instead of widening_optab_handler.
9804 * expr.c (expand_expr_real_2): Update calls to
9805 find_widening_optab_handler.
9806 * optabs.c (expand_widen_pattern_expr): Likewise.
9807 (expand_binop_directly): Take the insn_code as a parameter.
9808 (expand_binop): Only call find_widening_optab_handler for
9809 conversion optabs; use optab_handler otherwise. Update calls
9810 to find_widening_optab_handler and expand_binop_directly.
9811 Use convert_optab_handler instead of widening_optab_handler.
9812 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
9813 find_widening_optab_handler and use scalar_mode rather than
9815 (convert_plusminus_to_widen): Likewise.
9817 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9818 Alan Hayward <alan.hayward@arm.com>
9819 David Sherwood <david.sherwood@arm.com>
9821 * machmode.h (fixed_size_mode): New class.
9822 * rtl.h (get_pool_mode): Return fixed_size_mode.
9823 * gengtype.c (main): Add fixed_size_mode.
9824 * target.def (get_raw_result_mode): Return a fixed_size_mode.
9825 (get_raw_arg_mode): Likewise.
9826 * doc/tm.texi: Regenerate.
9827 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
9828 * targhooks.c (default_get_reg_raw_mode): Likewise.
9829 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
9830 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
9831 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
9832 (msp430_get_raw_result_mode): Likewise.
9833 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
9834 * dbxout.c (dbxout_parms): Require fixed-size modes.
9835 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
9836 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
9837 * omp-low.c (lower_oacc_reductions): Likewise.
9838 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
9839 (simplify_subreg): Update accordingly.
9840 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
9841 (force_const_mem): Update accordingly. Return NULL_RTX for modes
9842 that aren't fixed-size.
9843 (get_pool_mode): Return a fixed_size_mode.
9844 (output_constant_pool_2): Take a fixed_size_mode.
9846 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9847 Alan Hayward <alan.hayward@arm.com>
9848 David Sherwood <david.sherwood@arm.com>
9850 * doc/rtl.texi (vec_series): Document.
9851 (const): Say that the operand can be a vec_series.
9852 * rtl.def (VEC_SERIES): New rtx code.
9853 * rtl.h (const_vec_series_p_1): Declare.
9854 (const_vec_series_p): New function.
9855 * emit-rtl.h (gen_const_vec_series): Declare.
9856 (gen_vec_series): Likewise.
9857 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
9858 (gen_vec_series): Likewise.
9859 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
9860 * simplify-rtx.c (simplify_unary_operation): Handle negations
9862 (simplify_binary_operation_series): New function.
9863 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
9864 (test_vector_ops_series): New function.
9865 (test_vector_ops): Call it.
9866 * config/powerpcspe/altivec.md (altivec_lvsl): Use
9867 gen_const_vec_series.
9868 (altivec_lvsr): Likewise.
9869 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
9871 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9872 Alan Hayward <alan.hayward@arm.com>
9873 David Sherwood <david.sherwood@arm.com>
9875 * doc/rtl.texi (const): Update description of address constants.
9876 Say that vector constants are allowed too.
9877 * common.md (E, F): Use CONSTANT_P instead of checking for
9879 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
9880 checking for CONST_VECTOR.
9881 * expmed.c (make_tree): Use build_vector_from_val for a CONST
9883 * expr.c (expand_expr_real_2): Check for vector modes instead
9884 of checking for CONST_VECTOR.
9885 * rtl.h (const_vec_p): New function.
9886 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
9887 (unwrap_const_vec_duplicate): Handle them here too.
9889 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9890 David Malcolm <dmalcolm@redhat.com>
9891 Alan Hayward <alan.hayward@arm.com>
9892 David Sherwood <david.sherwood@arm.com>
9894 * rtl.h (vec_duplicate_p): New function.
9895 * selftest-rtl.c (assert_rtx_eq_at): New function.
9896 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
9897 (assert_rtx_eq_at): Declare.
9898 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
9899 * selftest-run-tests.c (selftest::run_tests): Call it.
9900 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
9901 (simplify_unary_operation_1): Recursively handle vector duplicates.
9902 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
9904 (simplify_subreg): Handle subregs of vector duplicates.
9905 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
9906 (selftest::simplify_rtx_c_tests): New functions.
9908 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9909 Alan Hayward <alan.hayward@arm.com>
9910 David Sherwood <david.sherwood@arm.com>
9912 * emit-rtl.h (gen_const_vec_duplicate): Declare.
9913 (gen_vec_duplicate): Likewise.
9914 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
9916 (gen_const_vector): ...here.
9917 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
9918 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
9919 whose elements are all equal.
9920 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
9921 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
9922 (simplify_relational_operation): Likewise.
9923 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
9925 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
9926 (aarch64_expand_vector_init): Likewise.
9927 * config/arm/arm.c (neon_vdup_constant): Likewise.
9928 (neon_expand_vector_init): Likewise.
9929 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
9930 (arm_block_set_unaligned_vect): Likewise.
9931 (arm_block_set_aligned_vect): Likewise.
9932 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
9933 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
9934 (expand_vec_perm_even_odd_pack): Likewise.
9935 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
9936 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
9937 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
9938 gen_const_vec_duplicate.
9939 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
9940 * config/mips/mips.c (mips_gen_const_int_vector): Use
9941 gen_const_vec_duplicate.
9942 (mips_expand_vector_init): Use CONST0_RTX.
9943 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
9944 (define_split): Use gen_const_vec_duplicate.
9945 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
9946 (define_split): Use gen_const_vec_duplicate.
9947 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
9948 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
9949 * config/spu/spu.c (spu_const): Likewise.
9951 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
9952 Alan Hayward <alan.hayward@arm.com>
9953 David Sherwood <david.sherwood@arm.com>
9955 * combine.c (can_change_dest_mode): Reject changes in
9956 REGMODE_NATURAL_SIZE.
9958 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
9960 * configure.ac (--enable-libssp): New.
9961 (gcc_cv_libc_provides_ssp): Check for explicit setting before
9962 trying to determine target-specific default. Adjust indentation.
9963 * configure: Regenerated.
9964 * doc/install.texi (Configuration): Expand --disable-libssp
9967 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
9969 config/i386/i386.c (ix86_expand_epilogue): Correct stack
9972 2017-10-31 Martin Jambor <mjambor@suse.cz>
9975 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
9977 2017-10-31 David Malcolm <dmalcolm@redhat.com>
9979 * auto-profile.c (autofdo_source_profile::read): Use
9980 UNKNOWN_LOCATION rather than 0.
9981 * diagnostic-core.h (warning_at_rich_loc): Rename to...
9982 (warning_at): ...this overload.
9983 (warning_at_rich_loc_n): Rename to...
9984 (warning_n): ...this overload.
9985 (error_at_rich_loc): Rename to...
9986 (error_at): ...this overload.
9987 (pedwarn_at_rich_loc): Rename to...
9988 (pedwarn): ...this overload.
9989 (permerror_at_rich_loc): Rename to...
9990 (permerror): ...this overload.
9991 (inform_at_rich_loc): Rename to...
9992 (inform): ...this overload.
9993 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
9994 (diagnostic_n_impl_richloc): Rename to...
9995 (diagnostic_n_impl): ...this rich_location *-based decl.
9996 (inform_at_rich_loc): Rename to...
9997 (inform): ...this, and add an assertion.
9998 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
9999 (warning_at_rich_loc): Rename to...
10000 (warning_at): ...this, and add an assertion.
10001 (warning_at_rich_loc_n): Rename to...
10002 (warning_n): ...this, and add an assertion.
10003 (warning_n): Update location_t-based implementation for removal of
10004 location_t-based diagnostic_n_impl.
10005 (pedwarn_at_rich_loc): Rename to...
10006 (pedwarn): ...this, and add an assertion.
10007 (permerror_at_rich_loc): Rename to...
10008 (permerror): ...this, and add an assertion.
10009 (error_n): Update for removal of location_t-based diagnostic_n_impl.
10010 (error_at_rich_loc): Rename to...
10011 (error_at): ...this, and add an assertion.
10012 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
10013 (driver::do_spec_on_infiles): Likewise.
10014 * substring-locations.c (format_warning_va): Update for renaming
10015 of inform_at_rich_loc.
10017 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
10019 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
10020 _Float<N>X built-in functions so that the variant without the
10021 "__builtin_" prefix is only enabled for the GNU C and Objective C
10022 languages when they are in non-strict ANSI/ISO mode.
10023 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
10024 * target.def (floatn_builtin_p): Add a target hook to control
10025 whether _Float<N> and _Float<N>X built-in functions without the
10026 "__builtin_" prefix are enabled, and return true for C and
10027 Objective C in the default hook. Include langhooks.h in
10029 * targhooks.h (default_floatn_builtin_p): Likewise.
10030 * targhooks.c (default_floatn_builtin_p): Likewise.
10031 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
10032 floatn_builtin_p target hook.
10033 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
10035 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
10036 Eric Botcazou <ebotcazou@adacore.com>
10038 PR rtl-optimization/81803
10039 * lra-constraints.c (curr_insn_transform): Also reload the whole
10040 register for a strict subreg no wider than a word if this is for
10041 a WORD_REGISTER_OPERATIONS target.
10043 2017-10-31 Jason Merrill <jason@redhat.com>
10045 * gdbinit.in: Skip over inlines from timevar.h.
10047 2017-10-31 Martin Liska <mliska@suse.cz>
10049 * doc/gcov.texi: Document new option.
10050 * gcov.c (print_usage): Likewise print it.
10051 (process_args): Support the argument.
10052 (format_count): New function.
10053 (format_gcov): Use the function.
10055 2017-10-31 Martin Liska <mliska@suse.cz>
10057 * gcov.c (struct name_map): do not use typedef.
10058 Define operator== and operator<.
10059 (name_search): Remove.
10060 (name_sort): Remove.
10061 (main): Do not allocate names.
10062 (process_file): Add vertical space.
10063 (generate_results): Use std::find.
10064 (release_structures): Do not release memory.
10065 (find_source): Use std::find.
10067 2017-10-31 Martin Liska <mliska@suse.cz>
10069 * gcov.c (struct line_info): Remove it's typedef.
10070 (line_info::line_info): Add proper ctor.
10071 (line_info::has_block): Do not use a typedef.
10072 (struct source_info): Do not use typedef.
10073 (circuit): Likewise.
10074 (get_cycles_count): Likewise.
10075 (output_intermediate_file): Iterate via vector iterator.
10076 (add_line_counts): Use std::vector methods.
10077 (accumulate_line_counts): Likewise.
10078 (output_lines): Likewise.
10080 2017-10-31 Martin Liska <mliska@suse.cz>
10082 * gcov.c (struct source_info): Remove typedef.
10083 (source_info::source_info): Add proper ctor.
10084 (accumulate_line_counts): Use struct, not it's typedef.
10085 (output_gcov_file): Likewise.
10086 (output_lines): Likewise.
10087 (main): Do not allocate an array.
10088 (output_intermediate_file): Use size of vector container.
10089 (process_file): Resize the vector.
10090 (generate_results): Do not preallocate, use newly added vector
10092 (release_structures): Do not release sources.
10093 (find_source): Use vector methods.
10094 (add_line_counts): Do not use typedef.
10096 2017-10-31 Martin Liska <mliska@suse.cz>
10098 * doc/gcov.texi: Document that.
10099 * gcov.c (add_line_counts): Mark lines with a non-executed
10101 (output_line_beginning): Handle such lines.
10102 (output_lines): Pass new argument.
10103 (output_intermediate_file): Print it in intermediate format.
10105 2017-10-31 Martin Liska <mliska@suse.cz>
10107 * color-macros.h: New file.
10108 * diagnostic-color.c: Factor out color related to macros to
10110 * doc/gcov.texi: Document -k option.
10111 * gcov.c (INCLUDE_STRING): Include string.h.
10112 (print_usage): Add -k option.
10113 (process_args): Parse it.
10114 (pad_count_string): New function.
10115 (output_line_beginning): Likewise.
10116 (DEFAULT_LINE_START): New macro.
10117 (output_lines): Support color output.
10119 2017-10-31 Martin Liska <mliska@suse.cz>
10121 PR gcov-profile/82633
10122 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
10123 their interaction with GCOV infrastructure.
10124 * configure.ac: Add -fkeep-{inline,static}-functions to
10126 * configure: Regenerate.
10128 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
10131 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
10133 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
10136 * config/rs6000/rs6000.md (allocate_stack): Force update interval
10137 into a register if it does not fit into an immediate offset field.
10139 2017-10-31 Olivier Hainque <hainque@adacore.com>
10141 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
10143 2017-10-31 Julia Koval <julia.koval@intel.com>
10145 * config.gcc: Add gfniintrin.h.
10146 * config/i386/gfniintrin.h: New.
10147 * config/i386/i386-builtin-types.def
10148 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
10149 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
10150 __builtin_ia32_vgf2p8affineinvqb_v32qi,
10151 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
10152 __builtin_ia32_vgf2p8affineinvqb_v16qi,
10153 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
10154 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
10155 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
10156 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
10157 V64QI_FTYPE_V64QI_V64QI_INT): New types.
10158 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
10159 * config/i386/immintrin.h: Include gfniintrin.h.
10160 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
10162 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
10164 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
10166 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10168 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
10169 (arm_ashldi3_1bit): Remove pattern.
10170 (ashrdi3): Remove shift by 1 expansion.
10171 (arm_ashrdi3_1bit): Remove pattern.
10172 (lshrdi3): Remove shift by 1 expansion.
10173 (arm_lshrdi3_1bit): Remove pattern.
10174 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
10175 cost of ashldi3 by 1.
10176 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
10177 (<shift>di3_neon): Likewise.
10179 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
10181 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
10182 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
10183 and (*aarch64_simd_mov<VQ:mode>).
10184 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
10185 pattern alternative.
10186 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
10187 attributes to match pattern alternative.
10189 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
10191 * config.gcc (powerpc*-*-*): Add emmintrin.h.
10192 * config/rs6000/emmintrin.h: New file.
10193 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
10195 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
10197 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
10198 * (movdi_vfp_cortexa8): Remove pattern.
10200 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10202 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
10204 (Specific, alpha*-dec-osf5.1): Remove.
10205 (Specific, mips-sgi-irix5): Remove.
10206 (Specific, mips-sgi-irix6): Remove.
10208 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10210 PR middle-end/22141
10211 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
10212 arguments to clear_bit_region_be.
10214 2017-10-30 Jim Wilson <wilson@tuliptree.org>
10216 * gimplify.c: Include memmodel.h.
10218 2017-10-30 Martin Jambor <mjambor@suse.cz>
10220 * omp-grid.c (grid_attempt_target_gridification): Also insert a
10221 condition whether loop should be executed at all.
10223 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
10225 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
10226 gimple folding of vec_madd() intrinsics.
10227 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
10228 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
10229 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
10231 2017-10-30 Richard Biener <rguenther@suse.de>
10233 PR tree-optimization/82762
10235 2017-10-23 Richard Biener <rguenther@suse.de>
10237 PR tree-optimization/82129
10239 2017-08-01 Richard Biener <rguenther@suse.de>
10241 PR tree-optimization/81181
10242 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10243 (compute_antic): ... end of iteration here.
10245 2017-10-30 Joseph Myers <joseph@codesourcery.com>
10247 * doc/invoke.texi (C Dialect Options): Document -std=c17,
10248 -std=iso9899:2017 and -std=gnu17.
10249 * doc/standards.texi (C Language): Document C17 support.
10250 * doc/cpp.texi (Overview): Mention -std=c17.
10251 (Standard Predefined Macros): Document C11 and C17 values of
10252 __STDC_VERSION__. Do not refer to C99 support as incomplete.
10253 * doc/extend.texi (Inline): Do not list individual options for
10254 standards newer than C99.
10255 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
10257 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
10260 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
10262 * asan.c (asan_finish_file): Align asan globals array by shadow
10265 2017-10-30 Jakub Jelinek <jakub@redhat.com>
10267 PR middle-end/22141
10268 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
10269 (struct store_immediate_info): Add bitregion_start and bitregion_end
10271 (store_immediate_info::store_immediate_info): Add brs and bre
10272 arguments and initialize bitregion_{start,end} from those.
10273 (struct merged_store_group): Add bitregion_start, bitregion_end,
10274 align_base and mask fields. Drop unnecessary struct keyword from
10275 struct store_immediate_info. Add do_merge method.
10276 (clear_bit_region_be): Use memset instead of loop storing zeros.
10277 (merged_store_group::do_merge): New method.
10278 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
10279 stores as long as the surrounding bitregions have no gaps.
10280 (merged_store_group::merge_overlapping): Use do_merge.
10281 (merged_store_group::apply_stores): Test that bitregion_{start,end}
10282 is byte aligned, rather than requiring that start and width are
10283 byte aligned. Drop unnecessary struct keyword from
10284 struct store_immediate_info. Allocate and populate also mask array.
10285 Make start of the arrays relative to bitregion_start rather than
10286 start and size them according to bitregion_{end,start} difference.
10287 (struct imm_store_chain_info): Drop unnecessary struct keyword from
10288 struct store_immediate_info.
10289 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
10290 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
10291 struct keyword from struct store_immediate_info.
10292 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
10293 between stores as long as the surrounding bitregions have no gaps.
10295 (struct split_store): Add orig non-static data member.
10296 (split_store::split_store): Initialize orig to false.
10297 (find_constituent_stmts): Return store_immediate_info *, non-NULL
10298 if there is exactly a single original stmt. Change stmts argument
10299 to pointer from reference, if NULL, don't push anything to it. Add
10300 first argument, use it to optimize skipping over orig stmts that
10301 are known to be before bitpos already. Simplify.
10302 (split_group): Return unsigned int count how many stores are or
10303 would be needed rather than a bool. Add allow_unaligned argument.
10304 Change split_stores argument from reference to pointer, if NULL,
10305 only do a dry run computing how many stores would be produced.
10306 Rewritten algorithm to use both alignment and misalign if
10307 !allow_unaligned and handle bitfield stores with gaps.
10308 (imm_store_chain_info::output_merged_store): Set start_byte_pos
10309 from bitregion_start instead of start. Compute allow_unaligned
10310 here, if true, do 2 split_group dry runs to compute which one
10311 produces fewer stores and prefer aligned if equal. Punt if
10312 new count is bigger or equal than original before emitting any
10313 statements, rather than during that. Remove no longer needed
10314 new_ssa_names tracking. Replace num_stmts with
10315 split_stores.length (). Use 32-bit stack allocated entries
10316 in split_stores auto_vec. Try to reuse original store lhs/rhs1
10317 if possible. Handle bitfields with gaps.
10318 (pass_store_merging::execute): Ignore bitsize == 0 stores.
10319 Compute bitregion_{start,end} for the stores and construct
10320 store_immediate_info with that. Formatting fixes.
10322 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
10325 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
10326 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
10328 2017-10-29 Jim Wilson <wilson@tuliptree.org>
10330 * gimplify.c: Include tm_p.h.
10332 * common.opt (gcoff): Re-add as ignored option.
10333 (gcoff1, gcoff2, gcoff3): Likewise.
10335 * Makefile.in (OBJS): Delete sdbout.o.
10336 (GTFILES): Delete $(srcdir)/sdbout.c.
10337 * debug.h: Delete sdb_debug_hooks.
10338 * final.c: Delete sdbout.h include.
10339 (final_scan_insn): Delete SDB_DEBUG check.
10340 (rest_of_clean_state): Likewise.
10341 * output.h: Delete sdb_begin_function_line.
10342 * sdbout.c: Delete.
10343 * sdbout.h: Delete.
10344 * toplev.c: Delete sdbout.h include.
10345 (process_options): Delete SDB_DEBUG check.
10346 * tree-core.h (tree_type_common): Delete pointer field of
10348 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
10349 TYPE_SYMTAB_POINTER.
10350 * tree.h (TYPE_SYMTAB_POINTER): Delete.
10351 (TYPE_SYMTAB_IS_POINTER): Delete.
10352 (TYPE_SYMTAB_IS_DIE): Renumber.
10353 * xcoffout.c: Refer to former sdbout.c file.
10354 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
10356 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
10357 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
10358 (Debugging Options): Delete -gcoff.
10359 (-gstabs): Delete SDB reference.
10361 (-gcoff@var{level}): Delete.
10362 * doc/passes.texi (Debugging information output): Delete SDB and
10363 sdbout.c references.
10364 * doc/tm.texi: Regenerate.
10365 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
10366 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
10368 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
10369 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
10370 (SDB_DEBUGGING_INFO): Delete.
10371 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
10372 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
10373 * target.def (output_source_filename): Delete COFF reference.
10375 * common.opt (gcoff): Delete.
10376 (gxcoff+): Update Negative chain.
10377 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
10379 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
10380 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
10381 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
10382 and SDB references.
10383 (expand_function_start): Change sdb reference to past tense.
10384 (expand_function_end): Change sdb reference to past tense.
10385 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
10386 * opts.c (debug_type_names): Delete coff entry.
10387 (common_handle_option): Delete OPT_gcoff case.
10388 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
10390 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
10391 * config/cris/cris.h: Delete SDB reference in comment.
10392 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
10393 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
10394 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
10395 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
10397 (ix86_expand_prologue): Likewise.
10398 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
10399 * config/ia64/ia64.h: Likewise.
10400 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
10401 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
10403 * config/mmix/mmix.h: Likewise.
10404 * config/nds32/nds32.c: Likewise.
10405 * config/stormy/storym16.h: Likewise.
10406 * config/visium/visium.h: Likewise.
10407 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
10409 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
10411 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
10412 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
10413 FRAME_POINTER_REGNUM point at high end of local var area.
10415 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
10417 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
10418 Move comment around. Do not reset best_edge for a copiable
10419 destination if the copy would cause a partition change.
10420 (better_edge_p): Remove redundant check.
10422 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10424 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
10426 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
10428 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
10429 for math functions that have _Float<N> and _Float<N>X variants.
10430 (mathfn_built_in_2): Add support for math functions that have
10431 _Float<N> and _Float<N>X variants.
10432 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
10433 (expand_builtin_mathfn_ternary): Add support for fma with
10434 _Float<N> and _Float<N>X variants.
10435 (expand_builtin): Likewise.
10436 (fold_builtin_3): Likewise.
10437 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
10438 create math function _Float<N> and _Float<N>X variants as external
10440 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
10441 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
10442 the __builtin_ prefix and if not strict ansi, without the prefix.
10443 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
10444 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
10445 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
10446 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
10447 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
10448 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
10449 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
10450 function signatures for fma _Float<N> and _Float<N>X variants.
10451 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
10452 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
10453 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
10454 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
10455 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
10456 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
10457 * gencfn-macros.c (print_case_cfn): Add support for math functions
10458 that have _Float<N> and _Float<N>X variants.
10459 (print_define_operator_list): Likewise.
10460 (fltfn_suffixes): Likewise.
10462 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
10463 for math functions that have _Float<N> and _Float<N>X variants.
10464 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
10465 and _Float<N>X variants.
10466 (COPYSIGN): Likewise.
10469 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
10470 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
10472 (integer_valued_read_call_p): Likewise.
10473 * fold-const-call.c (fold_const_call_ss): Likewise.
10474 (fold_const_call_sss): Add support for copysign, fmin, and fmax
10475 _Float<N> and _Float<N>X variants.
10476 (fold_const_call_ssss): Add support for fma _Float<N> and
10477 _Float<N>X variants.
10478 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
10479 support for copysign and fma _Float<N> and _Float<N>X variants.
10480 (backprop::process_builtin_call_use): Likewise.
10481 * tree-call-cdce.c (can_test_argument_range); Add support for
10482 sqrt _Float<N> and _Float<N>X variants.
10483 (edom_only_function): Likewise.
10484 (get_no_error_domain): Likewise.
10485 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
10486 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
10487 copysign _Float<N> and _Float<N>X variants.
10488 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
10489 handled by machine independent code.
10490 (FMAF128): Likewise.
10491 * doc/cpp.texi (Common Predefined Macros): Document defining
10492 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
10493 fma _Float<N> and _Float<N>X variants.
10495 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
10498 * config/i386/i386-modes.def (CCFPU): Remove definition.
10499 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
10500 (ix86_cc_modes_compatible): Ditto.
10501 (ix86_expand_carry_flag_compare): Ditto.
10502 (ix86_expand_int_movcc): Ditto.
10503 (ix86_expand_int_addcc): Ditto.
10504 (ix86_reverse_condition): Ditto.
10505 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
10506 Return true/false for unordered/ordered fp comparisons.
10507 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
10508 (ix86_prepare_fp_compare_args): Update for rename.
10509 (ix86_expand_fp_compare): Update for rename. Generate unordered
10510 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
10511 (ix86_expand_sse_compare_and_jump): Ditto.
10512 * config/i386/predicates.md (fcmov_comparison_operator):
10513 Remove CCFPU mode handling.
10514 (ix86_comparison_operator): Ditto.
10515 (ix86_carry_flag_operator): Ditto.
10516 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
10517 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
10518 (*cmpu<mode>_cc_i387): Ditto.
10519 (FPCMP): Remove mode iterator.
10520 (unord): Remove mode attribute.
10521 (unord_subst): New define_subst transformation
10522 (unord): New define_subst attribute.
10523 (unordered): Ditto.
10524 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
10525 (*cmpi<unord>xf_i387): Ditto.
10526 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
10527 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
10528 using unord_subst transformation.
10529 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
10530 (round_saeonly): Also handle CCFP mode.
10531 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
10532 Remove UNSPEC_SAHF unspec handling.
10534 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10536 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
10538 2017-10-27 Jeff Law <law@redhat.com>
10540 * gimple-ssa-sprintf.c: Include domwalk.h.
10541 (class sprintf_dom_walker): New class, derived from dom_walker.
10542 (sprintf_dom_walker::before_dom_children): New function.
10543 (struct call_info): Moved into sprintf_dom_walker class
10544 (compute_formath_length, handle_gimple_call): Likewise.
10545 (sprintf_length::execute): Call the dominator walker rather
10546 than walking the statements.
10548 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
10549 gimple statement locations.
10550 (check_array_bounds): Corresponding changes. Get the statement's
10551 location directly from wi->stmt.
10553 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
10556 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
10558 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
10560 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
10561 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
10563 2017-10-27 Jakub Jelinek <jakub@redhat.com>
10566 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
10567 * config/i386/i386.c (maybe_get_pool_constant): Removed.
10568 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
10569 maybe_get_pool_constant.
10570 * config/i386/predicates.md (zero_extended_scalar_load_operand):
10573 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10575 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
10576 2.26 caveat. Update gas and gld versions.
10577 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
10580 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10582 * cgraph.h (set_malloc_flag): Declare.
10583 * cgraph.c (set_malloc_flag_1): New function.
10584 (set_malloc_flag): Likewise.
10585 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
10586 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
10588 (read_ipa_call_summary): Add support for reading is_return_callee.
10589 (write_ipa_call_summary): Stream is_return_callee.
10590 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
10591 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
10592 ipa-prop.h, ipa-fnsummary.h.
10593 (pure_const_names): Change to static.
10594 (malloc_state_e): Define.
10595 (malloc_state_names): Define.
10596 (funct_state_d): Add field malloc_state.
10597 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
10598 (check_retval_uses): New function.
10599 (malloc_candidate_p): Likewise.
10600 (analyze_function): Add support for malloc attribute.
10601 (pure_const_write_summary): Stream malloc_state.
10602 (pure_const_read_summary): Add support for reading malloc_state.
10603 (dump_malloc_lattice): New function.
10604 (propagate_malloc): New function.
10605 (warn_function_malloc): New function.
10606 (ipa_pure_const::execute): Call propagate_malloc and
10607 ipa_free_fn_summary.
10608 (pass_local_pure_const::execute): Add support for malloc attribute.
10609 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
10610 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
10612 2017-10-27 Martin Liska <mliska@suse.cz>
10614 PR gcov-profile/82457
10615 * doc/invoke.texi: Document that one needs a non-strict ISO mode
10616 for fork-like functions to be properly instrumented.
10618 2017-10-27 Richard Biener <rguenther@suse.de>
10620 PR middle-end/81659
10621 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
10622 info when we redirected EH.
10624 2017-10-26 Michael Collison <michael.collison@arm.com>
10626 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
10628 (<optab>_trunchf<GPI:mode>2: New pattern.
10629 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
10630 * config/aarch64/iterators.md (wv): New mode attribute.
10631 (vf, VF): New mode attributes.
10632 (vgp, VGP): New mode attributes.
10633 (s): Update attribute with SImode and DImode prefixes.
10635 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10637 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
10638 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
10639 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
10640 (nios2_option_overide): Initialize it. Don't allow R0-relative
10641 addressing with PIC.
10642 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
10643 (nios2_symbolic_constant_p): Likewise.
10644 (nios2_legitimate_address_p): Likewise.
10645 (nios2_r0rel_section_name_p): New.
10646 (nios2_symbol_ref_in_r0rel_data_p): New.
10647 (nios2_emit_move_sequence): Handle r0rel_constant_p.
10648 (r0rel_constant_p): New.
10649 (nios2_print_operand_address): Handle r0rel_constant_p.
10650 (nios2_cdx_narrow_form_p): Likewise.
10651 * config/nios2/nios2.opt (mr0rel-sec=): New option.
10652 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
10653 (Nios II Options): Document -mr0rel-sec.
10655 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
10657 * config/nios2/nios2.c: Include xregex.h.
10658 (nios2_gprel_sec_regex): New.
10659 (nios2_option_overide): Initialize it. Don't allow GP-relative
10660 addressing with PIC.
10661 (nios2_small_section_name_p): Check for regex match.
10662 * config/nios2/nios2.opt (mgprel-sec=): New option.
10663 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
10664 (Nios II Options): Document -mgprel-sec.
10666 2017-10-26 Jim Wilson <wilson@tuliptree.org>
10668 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
10670 2017-10-26 Tom de Vries <tom@codesourcery.com>
10672 PR tree-optimization/82707
10673 * gimple.c (gimple_copy): Fix unsharing of
10674 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
10676 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
10678 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
10679 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
10680 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
10681 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
10682 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
10683 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
10684 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
10685 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
10686 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
10687 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
10688 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
10689 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
10690 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
10691 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
10692 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
10693 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
10694 _mm512_mask_cmpunord_ps_mask): New intrinsics.
10696 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
10698 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
10700 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
10701 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
10702 warning to rs6000.c. Remove the Undocumented flag, since it has
10704 (-mabi=ibmlongdouble): Likewise.
10705 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
10706 already set, set the default format for long double.
10707 (rs6000_debug_reg_global): Print whether long double is IBM or
10709 (rs6000_option_override_internal): Rework setting long double
10710 format. Only warn if the user is changing the long double default
10711 and they did not use -Wno-psabi.
10712 * doc/invoke.texi (PowerPC options): Update the documentation for
10713 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
10715 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10716 Alan Hayward <alan.hayward@arm.com>
10717 David Sherwood <david.sherwood@arm.com>
10719 * rtl.h (wider_subreg_mode): New function.
10720 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
10721 rather than an unsigned int *.
10722 * ira-color.c (regno_max_ref_width): Replace with...
10723 (regno_max_ref_mode): ...this new variable.
10724 (coalesced_pseudo_reg_slot_compare): Update accordingly.
10725 Use wider_subreg_mode.
10726 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
10727 rather than an unsigned int *.
10728 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
10729 (process_alt_operands): Likewise.
10730 (invariant_p): Likewise.
10731 * lra-spills.c (assign_mem_slot): Likewise.
10732 (add_pseudo_to_slot): Likewise.
10733 * lra.c (collect_non_operand_hard_regs): Likewise.
10734 (add_regs_to_insn_regno_info): Likewise.
10735 * reload1.c (regno_max_ref_width): Replace with...
10736 (regno_max_ref_mode): ...this new variable.
10737 (reload): Update accordingly. Update call to
10738 ira_sort_regnos_for_alter_reg.
10739 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
10740 (init_eliminable_invariants): Update to use regno_max_ref_mode.
10741 (scan_paradoxical_subregs): Likewise.
10743 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10745 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
10746 (aarch64_frame): Add emit_frame_chain boolean.
10747 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10748 Move eh_return case to aarch64_layout_frame.
10749 (aarch64_layout_frame): Initialize emit_frame_chain.
10750 (aarch64_expand_prologue): Use emit_frame_chain.
10752 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10754 * config/aarch64/aarch64.c (aarch64_layout_frame):
10755 Ensure LR is always stored at the bottom of the callee-saves.
10756 Remove rarely used frame layout which saves callee-saves at top of
10757 frame, so the store of LR can be used as a valid probe in all cases.
10759 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
10761 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
10762 Improve unaligned TImode/TFmode base/offset split.
10764 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10765 Alan Hayward <alan.hayward@arm.com>
10766 David Sherwood <david.sherwood@arm.com>
10768 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
10769 * combine.c (find_single_use_1): Likewise.
10770 (expand_field_assignment): Likewise.
10771 (move_deaths): Likewise.
10772 * lra-constraints.c (simplify_operand_subreg): Likewise.
10773 (curr_insn_transform): Likewise.
10774 * lra.c (collect_non_operand_hard_regs): Likewise.
10775 (add_regs_to_insn_regno_info): Likewise.
10776 * rtlanal.c (reg_referenced_p): Likewise.
10777 (covers_regno_no_parallel_p): Likewise.
10779 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10781 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
10782 Don't print any bits outside the precision of the value.
10783 * wide-int.cc (test_printing): Add some new tests.
10785 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10787 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
10788 supports -xbrace_comment option.
10789 * configure: Regenerate.
10790 * config.in: Regenerate.
10791 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
10792 (ASM_CPU_SPEC): Use it.
10794 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
10796 * target.def (static_rtx_alignment): New hook.
10797 * targhooks.h (default_static_rtx_alignment): Declare.
10798 * targhooks.c (default_static_rtx_alignment): New function.
10799 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
10800 * doc/tm.texi: Regenerate.
10801 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
10802 instead of targetm.constant_alignment. Remove call to
10803 set_mem_attributes.
10804 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10805 (cris_preferred_mininum_alignment): New function, split out from...
10806 (cris_constant_alignment): ...here.
10807 (cris_static_rtx_alignment): New function.
10808 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
10810 (ix86_constant_alignment): ...here.
10811 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10812 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10813 (mmix_static_rtx_alignment): New function.
10814 * config/spu/spu.c (spu_static_rtx_alignment): New function.
10815 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
10817 2017-10-26 Tamar Christina <tamar.christina@arm.com>
10820 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
10821 Add flag_trapping_math and flag_fp_int_builtin_inexact.
10823 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
10825 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
10826 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
10827 mark as a sign-extending load.
10828 (local_pic_load_u): Define.
10830 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
10832 PR middle-end/82062
10833 * fold-const.c (operand_equal_for_comparison_p): Also return true
10834 if ARG0 is a simple variant of ARG1 with narrower precision.
10835 (fold_ternary_loc): Always pass unstripped operands to the predicate.
10837 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
10839 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
10841 * i386.h (processor_costs): Add gather_static, gather_per_elt,
10842 scatter_static, scatter_per_elt.
10843 * x86-tune-costs.h: Add new cost entries.
10845 2017-10-25 Richard Biener <rguenther@suse.de>
10847 * tree-ssa-sccvn.h (vn_eliminate): Declare.
10848 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
10849 class pass_fre): Move to ...
10850 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
10851 class pass_fre): ... here and adjust for statistics.
10853 2017-10-25 Jakub Jelinek <jakub@redhat.com>
10856 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
10857 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
10858 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
10859 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
10862 2017-10-25 Richard Biener <rguenther@suse.de>
10864 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
10865 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
10866 eliminate_push_avail, eliminate_insert): Move inside...
10867 (class eliminate_dom_walker): ... this class in preparation
10869 (fini_eliminate): Remove by merging with ...
10870 (eliminate): ... this function. Adjust for class changes.
10871 (pass_pre::execute): Remove fini_eliminate call.
10872 (pass_fre::execute): Likewise.
10874 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10877 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
10878 (VPERMI2, VPERMI2I): New mode iterators.
10879 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
10880 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
10882 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
10883 mode iterator. Remove 3 old define_insn patterns.
10884 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
10885 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
10886 VPERMI2 mode iterator, remove the other two expanders.
10887 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
10888 to use VPERMI2 mode iterator, add another alternative for vpermi2*
10889 instructions, remove the other two patterns.
10890 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
10891 mode iterator, remove the other two patterns.
10892 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
10893 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
10894 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
10895 and adjust argument order accordingly.
10896 (ix86_expand_vec_perm): Adjust caller.
10897 (expand_vec_perm_1): Likewise.
10898 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
10899 (expand_vec_perm_vpermt2_vpshub2): ... this.
10900 (ix86_expand_vec_perm_const_1): Adjust caller.
10901 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
10904 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
10905 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
10906 (vec_shl_<mode>): Remove unused expander.
10907 (avx512bw_<shift_insn><mode>3): New define_insn.
10908 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
10909 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
10911 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
10914 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
10917 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10919 PR rtl-optimization/82396
10920 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
10921 (autopref_multipass_init): Simplify initialization.
10922 (autopref_rank_data): Simplify sort order.
10923 * gcc/sched-int.h (autopref_multipass_data_): Remove
10924 multi_mem_insn_p, min_offset and max_offset.
10926 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
10928 PR middle-end/60580
10929 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
10930 Check special value of flag_omit_frame_pointer.
10931 (aarch64_can_eliminate): Likewise.
10932 (aarch64_override_options_after_change_1): Simplify handling of
10933 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
10935 2017-10-24 Richard Biener <rguenther@suse.de>
10937 PR tree-optimization/82697
10938 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
10939 zero for conditional load and unconditional store.
10941 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10943 * doc/install.texi: Document bootstrap-cet.
10945 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
10948 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
10949 ENDBR instruction at function entrance if function is only
10952 2017-10-24 Jakub Jelinek <jakub@redhat.com>
10955 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
10956 patterns to better describe from which operation the CF is computed.
10957 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
10958 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
10959 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
10960 is 0, use _0 suffixed expanders instead of emitting a comparison
10963 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
10965 * config/i386/i386.md(*movsf_internal, *movdf_internal):
10966 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
10968 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
10970 PR middle-end/82569
10971 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
10972 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
10973 * loop-iv.c (iv_get_reaching_def): Likewise.
10974 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
10975 variable is promoted and the partition contains undefined values.
10977 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10979 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
10981 (nios2_address_cost): Define.
10982 (nios2_legitimize_address): Recognize (exp + constant) directly.
10983 (TARGET_ADDRESS_COST): Define.
10985 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
10987 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
10988 (nios2_symbolic_memory_operand_p): Declare.
10989 (nios2_split_large_constant): Declare.
10990 (nios2_split_symbolic_memory_operand): Declare.
10991 * config/nios2/nios2.c: Adjust includes.
10992 (nios2_symbolic_constant_allowed): New.
10993 (nios2_symbolic_constant_p): New.
10994 (nios2_plus_symbolic_constant_p): New.
10995 (nios2_valid_addr_expr_p): Recognize addresses involving
10996 symbolic constants.
10997 (nios2_legitimate_address_p): Likewise, also LO_SUM.
10998 (nios2_symbolic_memory_operand_p): New.
10999 (nios2_large_constant_p): New.
11000 (nios2_split_large_constant): New.
11001 (nios2_split_plus_large_constant): New.
11002 (nios2_split_symbolic_memory_operand): New.
11003 (nios2_legitimize_address): Code refactoring. Handle addresses
11004 involving symbolic constants.
11005 (nios2_emit_move_sequence): Likewise.
11006 (nios2_print_operand): Improve error output.
11007 (nios2_print_operand_address): Handle LO_SUM.
11008 (nios2_cdx_narrow_form_p): Likewise.
11009 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
11010 operands involving symbolic constants.
11011 (movhi_internal, movsi_internal): Likewise.
11012 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
11013 (extendhisi2, extendqi<mode>2): Likewise.
11015 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11017 * tree-pass.h (PROP_rtl_split_insns): Define.
11018 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
11020 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
11022 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
11024 2017-10-23 Jakub Jelinek <jakub@redhat.com>
11027 * target.def (const_not_ok_for_debug_p): Default to
11028 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
11029 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
11030 * targhooks.c (default_const_not_ok_for_debug_p): New function.
11031 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
11032 which targetm.const_not_ok_for_debug_p returned true.
11033 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
11035 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
11037 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
11038 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
11039 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
11041 (ix86_const_not_ok_for_debug_p): New function.
11042 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
11043 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
11045 2017-10-23 David Malcolm <dmalcolm@redhat.com>
11048 * system.h: Conditionally include "unique-ptr.h" if
11049 INCLUDE_UNIQUE_PTR is defined.
11050 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
11051 of defining INCLUDE_UNIQUE_PTR before including "system.h".
11053 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
11055 * config/rl78/rl78.md: New define_expand "subdi3".
11057 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
11060 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
11061 DF_REF_INSN if DF_REF_INSN_INFO is false.
11063 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
11065 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
11066 xmm_move instead of sse_move.
11067 (sse_store_index): New function.
11068 (ix86_register_move_cost): Be more sensible about mismatch stall;
11069 model AVX moves correctly; make difference between sse->integer and
11071 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
11072 moves; make difference between SSE and AVX.
11073 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
11074 and zmm_move. Increase size of sse load and store tables;
11075 add unaligned load and store tables; add ssemmx_to_integer.
11076 * x86-tune-costs.h: Update all entries according to real
11077 move latencies from Agner Fog's manual and chip documentation.
11079 2017-10-23 Jakub Jelinek <jakub@redhat.com>
11082 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
11083 * config/i386/constraints.md (Wf): New constraint.
11084 * config/i386/i386.md (UNSPEC_SBB): New unspec.
11085 (cmp<dwi>_doubleword): Removed.
11086 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
11087 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
11088 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
11089 expand with cmp<dwi>_doubleword. For LTU and GEU use
11090 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
11092 * common.opt (gcolumn-info): Enable by default.
11093 * doc/invoke.texi (gcolumn-info): Document new default.
11095 2017-10-23 Richard Biener <rguenther@suse.de>
11097 PR tree-optimization/82672
11098 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
11099 Fold the stmt if we propagated into it.
11101 2017-10-23 Richard Biener <rguenther@suse.de>
11103 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
11104 (bitmap_remove_expr_from_set): ... this. All callers call this
11105 for non-constant values.
11106 (bitmap_set_subtract): Rename to...
11107 (bitmap_set_subtract_expressions): ... this. Adjust and
11109 (bitmap_set_contains_value): Remove superfluous check.
11110 (bitmap_set_replace_value): Inline into single caller ...
11111 (bitmap_value_replace_in_set): ... here and simplify.
11112 (dependent_clean): Merge into ...
11113 (clean): ... this using an overload. Adjust.
11114 (prune_clobbered_mems): Adjust.
11115 (compute_antic_aux): Likewise.
11116 (compute_partial_antic_aux): Likewise.
11118 2017-10-23 Richard Biener <rguenther@suse.de>
11120 PR tree-optimization/82129
11122 2017-08-01 Richard Biener <rguenther@suse.de>
11124 PR tree-optimization/81181
11125 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
11126 (compute_antic): ... end of iteration here.
11128 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11130 * target.def (starting_frame_offset): New hook.
11131 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
11132 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
11133 * doc/tm.texi.in: Regenerate.
11134 * hooks.h (hook_hwi_void_0): Declare.
11135 * hooks.c (hook_hwi_void_0): New function.
11136 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
11137 STARTING_FRAME_OFFSET.
11138 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
11139 * reload1.c (reload): Likewise.
11140 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
11141 instead of STARTING_FRAME_OFFSET.
11142 * function.c (try_fit_stack_local): Likewise.
11143 (assign_stack_local_1): Likewise
11144 (instantiate_virtual_regs): Likewise.
11145 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
11146 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
11147 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
11148 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
11149 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
11150 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
11151 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
11152 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
11153 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
11154 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
11155 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
11156 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
11157 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
11158 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
11159 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
11160 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
11161 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
11162 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
11163 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
11164 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
11165 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
11166 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
11167 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
11168 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
11169 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
11170 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
11171 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
11172 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
11173 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
11174 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
11175 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
11176 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
11177 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
11178 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
11179 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
11180 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
11181 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
11182 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
11183 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
11184 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
11185 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
11186 * config/avr/avr.c (avr_starting_frame_offset): Make static and
11187 return a HOST_WIDE_INT.
11188 (avr_builtin_setjmp_frame_value): Use it instead of
11189 STARTING_FRAME_OFFSET.
11190 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11191 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
11192 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
11194 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11195 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
11196 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
11197 (TARGET_CONSTANT_ALIGNMENT): Redefine.
11198 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
11199 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
11200 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11201 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
11202 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
11203 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11204 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
11205 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
11207 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11208 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
11209 * config/mips/mips.c (mips_compute_frame_info): Refer to
11210 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11211 (mips_starting_frame_offset): New function.
11212 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11213 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
11214 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
11215 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
11216 and return a HOST_WIDE_INT.
11217 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11218 (mmix_initial_elimination_offset): Refer to
11219 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
11220 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
11221 * config/pa/pa.c (pa_starting_frame_offset): New function.
11222 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
11223 (pa_expand_prologue): Likewise.
11224 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11225 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
11226 !FRAME_GROWS_DOWNWARD handling to...
11227 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11228 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
11229 !FRAME_GROWS_DOWNWARD handling to...
11230 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11231 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
11232 !FRAME_GROWS_DOWNWARD handling to...
11233 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11234 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
11236 (rs6000_starting_frame_offset): New function.
11237 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
11238 !FRAME_GROWS_DOWNWARD handling to...
11239 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11240 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
11241 !FRAME_GROWS_DOWNWARD handling to...
11242 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11243 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
11244 !FRAME_GROWS_DOWNWARD handling to...
11245 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
11246 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
11247 (rs6000_starting_frame_offset): New function.
11248 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
11249 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
11250 * config/vax/vax.c (vax_starting_frame_offset): New function.
11251 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
11252 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11253 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
11254 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
11255 (TARGET_STARTING_FRAME_OFFSET): Redefine.
11256 * system.h (STARTING_FRAME_OFFSET): Poison.
11258 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11260 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
11261 SCALAR_TYPE_MODE instead of TYPE_MODE.
11263 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11264 Alan Hayward <alan.hayward@arm.com>
11265 David Sherwood <david.sherwood@arm.com>
11267 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
11269 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11270 Alan Hayward <alan.hayward@arm.com>
11271 David Sherwood <david.sherwood@arm.com>
11273 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
11275 2017-10-23 Richard Biener <rguenther@suse.de>
11277 PR tree-optimization/82129
11278 * tree-ssa-pre.c (bitmap_set_and): Remove.
11279 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
11280 canonicalizing expressions in the set to those with lowest
11281 ID rather than taking that from the first edge.
11283 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
11285 * combine.c (rtx_equal_for_field_assignment_p): Use
11286 byte_lowpart_offset.
11288 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11289 Alan Hayward <alan.hayward@arm.com>
11290 David Sherwood <david.sherwood@arm.com>
11292 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
11293 to a SUBREG_PROMOTED_VAR.
11295 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11296 Alan Hayward <alan.hayward@arm.com>
11297 David Sherwood <david.sherwood@arm.com>
11299 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
11300 (expand_debug_source_expr): Likewise.
11301 * combine.c (combine_simplify_rtx): Likewise.
11302 * cse.c (fold_rtx): Likewise.
11303 * optabs.c (expand_float): Likewise.
11304 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11305 (simplify_binary_operation_1): Likewise.
11307 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11308 Alan Hayward <alan.hayward@arm.com>
11309 David Sherwood <david.sherwood@arm.com>
11311 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
11312 (record_promoted_value): Likewise.
11313 * expr.c (expand_expr_real_2): Likewise.
11314 * ree.c (update_reg_equal_equiv_notes): Likewise.
11315 (combine_set_extension): Likewise.
11316 * rtlanal.c (low_bitmask_len): Likewise.
11317 * simplify-rtx.c (neg_const_int): Likewise.
11318 (simplify_binary_operation_1): Likewise.
11320 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11321 Alan Hayward <alan.hayward@arm.com>
11322 David Sherwood <david.sherwood@arm.com>
11324 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
11325 * regcprop.c (maybe_mode_change): Likewise.
11326 * reload1.c (alter_reg): Likewise.
11328 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11330 * inchash.h (inchash::hash::add_wide_int): New function.
11331 * lto-streamer-out.c (hash_tree): Use it.
11333 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
11335 * inchash.h (inchash::hash::add_wide_int): Rename to...
11336 (inchash::hash::add_hwi): ...this.
11337 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
11338 (polymorphic_call_target_hasher::hash): Likewise.
11339 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
11340 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
11341 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
11342 * lto-streamer-out.c (hash_tree): Likewise.
11343 * optc-save-gen.awk: Likewise.
11344 * tree.c (add_expr): Likewise.
11346 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11349 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
11350 for ordered inequality comparisons even with TARGET_IEEE_FP.
11352 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
11355 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
11356 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11357 Expand with cmp<dwi>_doubleword.
11359 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11361 * extend.texi: Add x86 specific to 'nocf_check' attribute.
11362 List CET intrinsics.
11363 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
11364 specific to -fcf-protection option.
11366 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11368 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
11369 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
11370 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
11371 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
11372 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
11373 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
11374 (extra_objs): Add cet.o for Linux/x86 targets.
11375 (tmake_file): Add i386/t-cet for Linux/x86 targets.
11376 * config/i386/cet.c: New file.
11377 * config/i386/cetintrin.h: Likewise.
11378 * config/i386/t-cet: Likewise.
11379 * config/i386/cpuid.h (bit_SHSTK): New.
11380 (bit_IBT): Likewise.
11381 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
11382 pass IBT and SHSTK bits.
11383 * config/i386/i386-builtin-types.def
11384 (VOID_FTYPE_UNSIGNED_PVOID): New.
11385 (VOID_FTYPE_UINT64_PVOID): Likewise.
11386 * config/i386/i386-builtin.def: Add CET intrinsics.
11387 * config/i386/i386-c.c (ix86_target_macros_internal): Add
11388 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
11389 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
11390 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
11392 * config/i386/i386.c (rest_of_insert_endbranch): New.
11393 (pass_data_insert_endbranch): Likewise.
11394 (pass_insert_endbranch): Likewise.
11395 (make_pass_insert_endbranch): Likewise.
11396 (ix86_notrack_prefixed_insn_p): Likewise.
11397 (ix86_target_string): Add -mibt, -mshstk flags.
11398 (ix86_option_override_internal): Add flag_cf_protection
11400 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
11401 (ix86_print_operand): Add 'notrack' prefix output.
11402 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
11403 (ix86_expand_builtin): Expand CET intrinsics.
11404 (x86_output_mi_thunk): Add 'endbranch' instruction.
11405 * config/i386/i386.h (TARGET_IBT): New.
11406 (TARGET_IBT_P): Likewise.
11407 (TARGET_SHSTK): Likewise.
11408 (TARGET_SHSTK_P): Likewise.
11409 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
11410 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
11411 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
11412 (builtin_setjmp_setup): New pattern.
11413 (builtin_longjmp): Likewise.
11414 (rdssp<mode>): Likewise.
11415 (incssp<mode>): Likewise.
11416 (saveprevssp): Likewise.
11417 (rstorssp): Likewise.
11418 (wrss<mode>): Likewise.
11419 (wruss<mode>): Likewise.
11420 (setssbsy): Likewise.
11421 (clrssbsy): Likewise.
11422 (nop_endbr): Likewise.
11423 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
11425 * config/i386/immintrin.h: Include <cetintrin.h>.
11426 * config/i386/linux-common.h
11427 (file_end_indicate_exec_stack_and_cet): New prototype.
11428 (TARGET_ASM_FILE_END): New.
11430 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11432 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
11433 latencies instead of having separate table; make difference between
11434 integer and float costs.
11435 * i386.h (processor_costs): Remove scalar_stmt_cost,
11436 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
11437 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
11439 * x86-tune-costs.h: Remove entries which has been removed in
11440 procesor_costs from all tables; make cond_taken_branch_cost
11441 and cond_not_taken_branch_cost COST_N_INSNS based.
11443 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
11445 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
11447 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11449 * config/i386/i386.md (isa): Remove fma_avx512f.
11450 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
11451 <avx512>_fmadd_<mode>_mask3<round_name>,
11452 <avx512>_fmsub_<mode>_mask<round_name>,
11453 <avx512>_fmsub_<mode>_mask3<round_name>,
11454 <avx512>_fnmadd_<mode>_mask<round_name>,
11455 <avx512>_fnmadd_<mode>_mask3<round_name>,
11456 <avx512>_fnmsub_<mode>_mask<round_name>,
11457 <avx512>_fnmsub_<mode>_mask3<round_name>,
11458 <avx512>_fmaddsub_<mode>_mask<round_name>,
11459 <avx512>_fmaddsub_<mode>_mask3<round_name>,
11460 <avx512>_fmsubadd_<mode>_mask<round_name>,
11461 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
11462 (*vec_widen_umult_even_v16si<mask_name>,
11463 *vec_widen_smult_even_v16si<mask_name>): Likewise.
11464 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
11466 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11468 * extend.texi: Add 'nocf_check' documentation.
11469 * gimple.texi: Add second parameter to
11470 gimple_build_call_from_tree.
11471 * invoke.texi: Add -fcf-protection documentation.
11472 * rtl.texi: Add REG_CALL_NOTRACK documenation.
11474 2017-10-20 Richard Biener <rguenther@suse.de>
11476 PR tree-optimization/82473
11477 * tree-vect-loop.c (vectorizable_reduction): Properly get at
11478 the largest input type.
11480 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
11482 * c-attribs.c (handle_nocf_check_attribute): New function.
11483 (c_common_attribute_table): Add 'nocf_check' handling.
11484 * gimple-parser.c: Add second argument NULL to
11485 gimple_build_call_from_tree.
11486 * attrib.c (comp_type_attributes): Check nocf_check attribute.
11487 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
11489 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
11490 * common.opt: Add fcf-protection flag.
11491 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
11492 * flag-types.h: Add enum cf_protection_level.
11493 * gimple.c (gimple_build_call_from_tree): Add second parameter.
11494 Add 'nocf_check' attribute propagation to gimple call.
11495 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
11496 (gimple_build_call_from_tree): Update prototype.
11497 (gimple_call_nocf_check_p): New function.
11498 (gimple_call_set_nocf_check): Likewise.
11499 * gimplify.c: Add second argument to gimple_build_call_from_tree.
11500 * ipa-icf.c: Add nocf_check attribute in statement hash.
11501 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
11502 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
11503 * toplev.c (process_options): Add flag_cf_protection handling.
11505 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11507 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
11509 2017-10-20 Richard Biener <rguenther@suse.de>
11511 PR tree-optimization/82603
11512 * tree-if-conv.c (predicate_mem_writes): Make sure to only
11513 remove false predicated stores.
11515 2017-10-20 Richard Biener <rguenther@suse.de>
11517 * graphite-isl-ast-to-gimple.c
11518 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
11519 Remove return value and simplify, dump copied stmt after lhs
11521 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
11522 Reduce dump verbosity.
11523 (gsi_insert_earliest): Likewise.
11524 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
11525 * graphite.c (print_global_statistics): Adjust dumping.
11526 (print_graphite_scop_statistics): Likewise.
11527 (print_graphite_statistics): Do not dump loops here.
11528 (graphite_transform_loops): But here.
11530 2017-10-20 Nicolas Roche <roche@adacore.com>
11532 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
11533 * configure: Regenerate.
11535 2017-10-20 Jakub Jelinek <jakub@redhat.com>
11538 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
11539 functions when optimizing replace GIMPLE_RETURN stmts with
11540 calls to __builtin_unreachable ().
11543 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
11544 for -fsanitize=thread link of executables.
11545 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
11546 link of executables.
11549 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
11550 New mode iterators.
11551 (<shift_insn><mode>3<mask_name>): Change the last of the 3
11552 define_insns for logical vector shifts to use VI248_AVX512BW
11553 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
11554 condition, useless isa and prefix attributes. Change the first
11555 2 of these define_insns to ...
11556 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
11557 define_insn for avx512vl.
11558 (<shift_insn><mode>3): ... and this, new define_insn without
11559 masking for non-avx512vl.
11562 * config/i386/sse.md (*andnot<mode>3,
11563 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
11564 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
11565 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
11566 not applied use empty suffix even for TARGET_AVX512VL.
11567 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
11568 is applied, supply evex,evex or evex,evex,evex instead of just
11571 2017-10-20 Julia Koval <julia.koval@intel.com>
11573 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
11574 (OPTION_MASK_ISA_GFNI_UNSET): New.
11575 (ix86_handle_option): Handle OPT_mgfni.
11576 * config/i386/cpuid.h (bit_GFNI): New.
11577 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
11578 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
11579 * config/i386/i386.c (ix86_target_string): Add -mgfni.
11580 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
11581 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
11582 * config/i386/i386.opt: Add mgfni.
11584 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
11586 * config/msp430/msp430.c (msp430_option_override): Disable
11587 -fdelete-null-pointer-checks.
11588 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
11590 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11592 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
11593 of x87 and SSE instructions.
11595 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
11597 * asan.c (create_cond_insert_point): Do not update edge count.
11598 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
11599 (afdo_propagate_circuit): Likewise.
11600 (afdo_calculate_branch_prob): Likewise.
11601 (afdo_annotate_cfg): Likewise.
11602 * basic-block.h (struct edge_def): Remove count.
11603 (edge_def::count): New accessor.
11604 * bb-reorder.c (rotate_loop): Update.
11605 (find_traces_1_round): Update.
11606 (connect_traces): Update.
11607 (sanitize_hot_paths): Update.
11608 * cfg.c (unchecked_make_edge): Update.
11609 (make_single_succ_edge): Update.
11610 (check_bb_profile): Update.
11611 (dump_edge_info): Update.
11612 (update_bb_profile_for_threading): Update.
11613 (scale_bbs_frequencies_int): Update.
11614 (scale_bbs_frequencies_gcov_type): Update.
11615 (scale_bbs_frequencies_profile_count): Update.
11616 (scale_bbs_frequencies): Update.
11617 * cfganal.c (connect_infinite_loops_to_exit): Update.
11618 * cfgbuild.c (compute_outgoing_frequencies): Update.
11619 (find_many_sub_basic_blocks): Update.
11620 * cfgcleanup.c (try_forward_edges): Update.
11621 (try_crossjump_to_edge): Update
11622 * cfgexpand.c (expand_gimple_cond): Update
11623 (expand_gimple_tailcall): Update
11624 (construct_exit_block): Update
11625 * cfghooks.c (verify_flow_info): Update
11626 (redirect_edge_succ_nodup): Update
11627 (split_edge): Update
11628 (make_forwarder_block): Update
11629 (duplicate_block): Update
11630 (account_profile_record): Update
11631 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11632 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11633 * cfgloopmanip.c (scale_loop_profile): Update.
11635 (lv_adjust_loop_entry_edge): Update.
11636 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
11637 (force_nonfallthru_and_redirect): Update.
11638 (purge_dead_edges): Update.
11639 (rtl_flow_call_edges_add): Update.
11640 * cgraphunit.c (init_lowered_empty_function): Update.
11641 (cgraph_node::expand_thunk): Update.
11642 * gimple-pretty-print.c (dump_probability): Update.
11643 (dump_edge_probability): Update.
11644 * gimple-ssa-isolate-paths.c (isolate_path): Update.
11645 * haifa-sched.c (sched_create_recovery_edges): Update.
11646 * hsa-gen.c (convert_switch_statements): Update.
11647 * ifcvt.c (dead_or_predicable): Update.
11648 * ipa-inline-transform.c (inline_transform): Update.
11649 * ipa-split.c (split_function): Update.
11650 * ipa-utils.c (ipa_merge_profiles): Update.
11651 * loop-doloop.c (add_test): Update.
11652 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
11653 * lto-streamer-in.c (input_cfg): Update.
11654 (input_function): Update.
11655 * lto-streamer-out.c (output_cfg): Update.
11656 * modulo-sched.c (sms_schedule): Update.
11657 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11658 * predict.c (maybe_hot_edge_p): Update.
11659 (unlikely_executed_edge_p): Update.
11660 (probably_never_executed_edge_p): Update.
11661 (dump_prediction): Update.
11662 (drop_profile): Update.
11663 (propagate_unlikely_bbs_forward): Update.
11664 (determine_unlikely_bbs): Update.
11665 (force_edge_cold): Update.
11666 * profile.c (compute_branch_probabilities): Update.
11667 * reg-stack.c (better_edge): Update.
11668 * shrink-wrap.c (handle_simple_exit): Update.
11669 * tracer.c (better_p): Update.
11670 * trans-mem.c (expand_transaction): Update.
11671 (split_bb_make_tm_edge): Update.
11672 * tree-call-cdce.c: Update.
11673 * tree-cfg.c (gimple_find_sub_bbs): Update.
11674 (gimple_split_edge): Update.
11675 (gimple_duplicate_sese_region): Update.
11676 (gimple_duplicate_sese_tail): Update.
11677 (gimple_flow_call_edges_add): Update.
11678 (insert_cond_bb): Update.
11679 (execute_fixup_cfg): Update.
11680 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
11681 * tree-complex.c (expand_complex_div_wide): Update.
11682 * tree-eh.c (lower_resx): Update.
11683 (unsplit_eh): Update.
11684 (cleanup_empty_eh_move_lp): Update.
11685 * tree-inline.c (copy_edges_for_bb): Update.
11686 (freqs_to_counts): Update.
11687 (copy_cfg_body): Update.
11688 * tree-ssa-dce.c (remove_dead_stmt): Update.
11689 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11690 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
11691 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
11692 (unloop_loops): Update.
11693 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11694 * tree-ssa-loop-split.c (connect_loops): Update.
11695 (split_loop): Update.
11696 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11697 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
11698 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
11699 * tree-ssa-reassoc.c (branch_fixup): Update.
11700 * tree-ssa-tail-merge.c (replace_block_by): Update.
11701 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
11702 (compute_path_counts): Update.
11703 (update_profile): Update.
11704 (recompute_probabilities): Update.
11705 (update_joiner_offpath_counts): Update.
11706 (estimated_freqs_path): Update.
11707 (freqs_to_counts_path): Update.
11708 (clear_counts_path): Update.
11709 (ssa_fix_duplicate_block_edges): Update.
11710 (duplicate_thread_path): Update.
11711 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
11712 (case_bit_test_cmp): Update.
11713 (collect_switch_conv_info): Update.
11714 (gen_inbound_check): Update.
11715 (do_jump_if_equal): Update.
11716 (emit_cmp_and_jump_insns): Update.
11717 * tree-tailcall.c (decrease_profile): Update.
11718 (eliminate_tail_call): Update.
11719 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11720 (vect_do_peeling): Update.
11721 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11722 * ubsan.c (ubsan_expand_null_ifn): Update.
11723 (ubsan_expand_ptr_ifn): Update.
11724 * value-prof.c (gimple_divmod_fixed_value): Update.
11725 (gimple_mod_pow2): Update.
11726 (gimple_mod_subtract): Update.
11727 (gimple_ic): Update.
11728 (gimple_stringop_fixed_value): Update.
11730 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11733 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
11735 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
11737 PR rtl-optimization/82395
11738 * ira-color.c (allocno_priority_compare_func): Fix comparison step
11739 based on non_spilled_static_chain_regno_p.
11741 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
11743 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
11744 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
11745 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
11747 2017-10-19 Martin Sebor <msebor@redhat.com>
11749 PR tree-optimization/82596
11750 * tree.c (array_at_struct_end_p): Handle STRING_CST.
11752 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11754 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
11755 (get_mem_refs_of_builtin_call): Likewise.
11756 * builtins.c (expand_builtin_apply): Adjust call to
11757 allocate_dynamic_stack_space.
11758 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
11759 the third argument to allocate_dynamic_stack_space, otherwise -1.
11760 (expand_builtin): Deal with all alloca variants.
11761 (is_inexpensive_builtin): Likewise.
11762 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
11763 * calls.c (special_function_p): Deal with all alloca variants.
11764 (initialize_argument_information): Adjust call to
11765 allocate_dynamic_stack_space.
11766 (expand_call): Likewise.
11767 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
11768 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
11769 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
11770 use it for the stack usage computation.
11771 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
11772 * function.c (gimplify_parameters): Call build_alloca_call_expr.
11773 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
11774 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
11775 (in_loop_p): Remove first argument and useless check.
11776 (pass_walloca::execute): Remove useless test and adjust call to above.
11777 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
11778 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
11779 (gimplify_call_expr): Deal with all alloca variants.
11780 * hsa-gen.c (gen_hsa_alloca): Likewise.
11781 (gen_hsa_insns_for_call): Likewise.
11782 * ipa-pure-const.c (special_builtin_state): Likewise.
11783 * tree-chkp.c (chkp_build_returned_bound): Likewise.
11784 * tree-object-size.c (alloc_object_size): Likewise.
11785 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
11786 (call_may_clobber_ref_p_1): Likewise.
11787 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
11788 (ccp_fold_stmt): Likewise.
11789 (optimize_stack_restore): Likewise.
11790 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
11791 (mark_all_reaching_defs_necessary_1): Likewise.
11792 (propagate_necessity): Likewise.
11793 (eliminate_unnecessary_stmts): Likewise.
11794 * tree.c (build_common_builtin_nodes): Build
11795 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
11796 (build_alloca_call_expr): New function.
11797 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
11798 (CASE_BUILT_IN_ALLOCA): Likewise.
11799 (build_alloca_call_expr): Declare.
11800 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
11802 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
11805 * dwarf2out.c (new_die_raw): New static inline function.
11806 (new_die): Use it to create the DIE.
11807 (add_AT_external_die_ref): Likewise.
11808 (clone_die): Likewise.
11809 (clone_as_declaration): Likewise.
11810 (dwarf2out_vms_debug_main_pointer): Likewise.
11811 (base_type_die): Likewise. Remove early return for corner cases.
11812 Do not call add_pubtype on the DIE here.
11813 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
11814 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
11815 typedefs for base types with DW_AT_endianity. Make sure a DIE with
11816 native order exists for base types, attach the DIE manually and call
11817 add_pubtype on it. Do not equate a reverse order DIE to the type.
11819 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
11821 * config/arm/arm.c (align_ok_ldrd_strd): New function.
11822 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
11824 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
11826 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11828 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
11829 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
11830 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
11831 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
11832 * opts.c (sanitizer_opts): Add builtin.
11833 * ubsan.c (instrument_builtin): New function.
11834 (pass_ubsan::execute): Call it.
11835 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
11836 * doc/invoke.texi: Document -fsanitize=builtin.
11838 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
11839 builtins, store max (log2 (align), 0) into uchar field instead of
11840 align into uptr field.
11841 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
11842 store uchar 0 field instead of uptr 0 field.
11843 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
11844 instead of passing one address of struct with 2 locations pass
11845 two addresses of structs with 1 location each.
11846 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
11847 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
11848 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
11849 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
11850 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
11851 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
11852 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
11853 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
11855 2017-10-19 Martin Liska <mliska@suse.cz>
11858 * file-find.c (remove_prefix): Remove.
11859 * file-find.h (remove_prefix): Likewise.
11860 * gcc-ar.c: Remove smartness of lookup.
11862 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
11864 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
11865 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
11866 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
11868 2017-10-19 Jakub Jelinek <jakub@redhat.com>
11871 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
11872 (setcc + and to xor + setcc): New peephole2.
11874 2017-10-19 Tom de Vries <tom@codesourcery.com>
11876 * doc/sourcebuild.texi (Test Directives, Variants of
11877 dg-require-support): Add dg-require-stack-size.
11879 2017-10-19 Martin Liska <mliska@suse.cz>
11882 * gimplify.c (gimplify_decl_expr): Do not instrument variables
11883 that have a large alignment.
11884 (gimplify_target_expr): Likewise.
11886 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
11888 PR rtl-optimization/82602
11889 * ira.c (rtx_moveable_p): Return false for volatile asm.
11891 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
11894 * config/i386/i386-modes.def (CCGZ): New CC mode.
11895 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
11896 * config/i386/predicates.md (ix86_comparison_operator):
11898 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
11899 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
11900 with double-word subtraction.
11901 (put_condition_code): Handle CCGZmode.
11903 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
11905 * wide-int.cc (debug (const wide_int &)): New.
11906 (debug (const wide_int *)): New.
11907 (debug (const widest_int &)): New.
11908 (debug (const widest_int *)): New.
11910 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
11912 PR middle-end/82556
11913 * lra-constraints.c (curr_insn_transform): Use non-input operand
11914 instead of output one for matched reload.
11916 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11918 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
11919 (tree-ssa-loop-ivopts.h): New header file.
11920 (struct builtin_info): New fields.
11921 (classify_builtin_1): Compute and record base and offset parts for
11922 memset builtin partition by calling strip_offset.
11923 (offset_cmp, fuse_memset_builtins): New functions.
11924 (finalize_partitions): Fuse adjacent memset partitions by calling
11926 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
11927 Expose the interface.
11928 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
11930 2017-10-18 Bin Cheng <bin.cheng@arm.com>
11932 PR tree-optimization/82574
11933 * tree-loop-distribution.c (find_single_drs): New parameter. Check
11934 that data reference must be executed exactly once per iteration
11935 against the outermost loop in nest.
11936 (classify_partition): Update call to above function.
11938 2017-10-18 Richard Biener <rguenther@suse.de>
11940 PR tree-optimization/82591
11941 * graphite.c (graphite_transform_loops): Move code gen message
11943 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
11944 Here. Handle scop_to_isl_ast failing.
11945 (scop_to_isl_ast): Limit the number of ISL operations.
11947 2017-10-18 Richard Biener <rguenther@suse.de>
11949 * graphite-isl-ast-to-gimple.c
11950 (translate_isl_ast_to_gimple::set_rename): Simplify.
11951 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
11952 (graphite_copy_stmts_from_block): ... here.
11953 (copy_bb_and_scalar_dependences): Simplify.
11954 (add_parameters_to_ivs_params): Canonicalize.
11955 (generate_entry_out_of_ssa_copies): Simplify.
11956 * graphite-sese-to-poly.c (extract_affine_name): Simplify
11957 by passing in ISL dimension.
11958 (parameter_index_in_region_1): Rename to ...
11959 (parameter_index_in_region): ... this.
11960 (extract_affine): Adjust assert, pass down parameter index.
11961 (add_param_constraints): Use range-info when available.
11962 (build_scop_context): Adjust.
11963 * sese.c (new_sese_info): Adjust.
11964 (free_sese_info): Likewise.
11965 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
11966 Remove unused typedefs.
11967 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
11969 2017-10-18 Martin Liska <mliska@suse.cz>
11971 * combine.c (simplify_compare_const): Add gcc_fallthrough.
11973 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11975 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
11976 (s390_sched_init): Do not reset s390_sched_state if we entered the
11977 current basic block via a fallthru edge and all others are unlikely.
11979 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11981 * config/s390/s390.c (NUM_SIDES): New variable.
11982 (LONGRUNNING_THRESHOLD): New variable.
11983 (LATENCY_FACTOR): New variable.
11984 (s390_sched_score): Decrease score for long-running instructions on
11986 (s390_sched_variable_issue): Perform bookkeeping for long-running
11989 2017-10-18 Richard Biener <rguenther@suse.de>
11991 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
11992 Simplify with removal of the parameter rename map.
11993 (set_rename): Likewise.
11994 (should_copy_to_new_region): Likewise.
11995 (graphite_copy_stmts_from_block): Likewise.
11996 (copy_bb_and_scalar_dependences): Remove initialization of
11997 unused copied_bb_map.
11998 (copy_def): Remove.
11999 (copy_internal_parameters): Likewise.
12000 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
12001 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
12002 Use INTEGRAL_TYPE_P.
12003 (parameter_index_in_region_1): Rename to ...
12004 (assign_parameter_index_in_region): ... this. Assert we have
12005 a parameter we handle.
12006 (scan_tree_for_params): Adjust.
12007 * sese.h (parameter_rename_map_t): Remove.
12008 (struct sese_info_t): Remove unused parameter_rename_map and
12009 copied_bb_map members.
12010 * sese.c (new_sese_info): Adjust.
12011 (free_sese_info): Likewise.
12013 2017-10-18 Martin Liska <mliska@suse.cz>
12016 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
12017 on an abnormal edge.
12019 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
12021 * doc/invoke.texi (ffunction-sections and fdata-sections):
12024 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
12026 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
12027 the use statement can throw internally.
12029 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
12031 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
12032 any RTX present on the RHS of a SET.
12033 * compare-elim.c (try_eliminate_compare): Restore comment.
12035 2017-10-17 Jakub Jelinek <jakub@redhat.com>
12037 * langhooks.h (struct lang_hooks): Document that tree_size langhook
12038 may be also called on tcc_type nodes.
12039 * langhooks.c (lhd_tree_size): Likewise.
12041 2017-10-17 David Malcolm <dmalcolm@redhat.com>
12043 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
12044 format_warning_at_substring.
12045 (maybe_warn): Convert source_range * param to a location_t. Pass
12046 UNKNOWN_LOCATION rather than NULL to fmtwarn.
12047 (format_directive): Remove code to extract source_ranges and
12048 source_range * in favor of just a location_t.
12049 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
12051 * substring-locations.c (format_warning_va): Convert
12052 source_range * param to a location_t.
12053 (format_warning_at_substring): Likewise.
12054 * substring-locations.h (format_warning_va): Likewise.
12055 (format_warning_at_substring): Likewise.
12057 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
12059 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
12061 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
12062 and scatter/gather ops.
12064 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
12065 vec_gather_load and vec_scatter_store.
12066 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
12067 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
12068 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
12069 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
12070 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
12071 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
12073 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
12075 * reg-stack.c (compare_for_stack_reg): Add bool argument.
12076 Detect FTST instruction and handle its register pops. Only pop
12077 second operand if can_pop_second_op is true.
12078 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
12079 set can_pop_second_op to false in the compare_for_stack_reg call.
12081 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
12082 output_fp_compare for stack register operands.
12083 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
12084 instructions here. Do not emit stack register pops here. Assert
12085 that FCOMPP pops next to top stack register. Rewrite function.
12087 2017-10-17 Nathan Sidwell <nathan@acm.org>
12089 PR middle-end/82577
12090 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
12091 use DECL_ASSEMBLER_NAME_RAW.
12093 PR middle-end/82546
12094 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
12097 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
12098 Wilco Dijkstra <wilco.dijkstra@arm.com>
12100 * builtins.c (expand_builtin_update_setjmp_buf): Add a
12101 converstion to Pmode from the buf_addr.
12103 2017-10-17 Richard Biener <rguenther@suse.de>
12105 * graphite-dependences.c (scop_get_reads_and_writes): Change
12106 output parameters to references.
12108 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
12110 PR 71026/tree-optimization
12111 * fold-const.c (distribute_real_division): Removed.
12112 (fold_binary_loc): Remove calls to distribute_real_divison.
12114 2017-10-17 Richard Biener <rguenther@suse.de>
12116 * graphite-scop-detection.c
12117 (scop_detection::stmt_has_simple_data_refs_p): Always use
12118 the full nest as region.
12119 (try_generate_gimple_bb): Likewise.
12120 * sese.c (scalar_evolution_in_region): Simplify now that
12121 SCEV can handle instantiation in regions.
12122 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
12123 in the non-loop part of a function if requested.
12125 2017-10-17 Richard Biener <rguenther@suse.de>
12127 PR tree-optimization/82563
12128 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
12130 (graphite_regenerate_ast_isl): Call it.
12131 * graphite-scop-detection.c (build_scops): Remove entry edge split.
12133 2017-10-17 Jakub Jelinek <jakub@redhat.com>
12135 PR tree-optimization/82549
12136 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
12137 Formatting fixes. Instead of calling make_bit_field_ref with negative
12140 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
12142 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
12143 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
12144 _mm_maskz_reduce_ss): New.
12145 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
12146 __builtin_ia32_reducess_mask): Ditto..
12147 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
12148 * config/i386/sse.md (reduces<mode>): Renamed to ...
12149 (reduces<mode><mask_scalar_name>): ... this.
12150 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
12152 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
12153 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
12155 2017-10-16 David Malcolm <dmalcolm@redhat.com>
12157 * Makefile.in (OBJS): Add unique-ptr-tests.o.
12158 * selftest-run-tests.c (selftest::run_tests): Call
12159 selftest::unique_ptr_tests_cc_tests.
12160 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
12161 * unique-ptr-tests.cc: New file.
12163 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
12166 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12168 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12169 (make_hard_regno_born, make_hard_regno_dead): Update
12170 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
12172 2017-10-16 Jeff Law <law@redhat.com>
12174 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12176 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
12178 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
12180 2017-10-16 Olivier Hainque <hainque@adacore.com>
12182 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
12183 with_cpu if we were configured for an e500v2 target cpu name.
12185 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
12187 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
12188 * doc/invoke.texi: Document +nodsp as a valid extension for
12191 2017-10-16 Martin Liska <mliska@suse.cz>
12193 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
12194 (test_set_range): Likewise.
12195 (test_range_functions): Rename to ...
12196 (test_bit_in_range): ... this.
12197 (sbitmap_c_tests): Add new test.
12199 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12201 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
12203 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
12204 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
12206 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12208 * config/aarch64/aarch64-builtins.c
12209 (aarch64_types_quadopu_lane_qualifiers): New.
12210 (TYPES_QUADOPU_LANE): New.
12211 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
12212 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
12213 (aarch64_<sur>dot_laneq<vsi2qi>): New.
12214 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
12215 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
12216 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
12217 (Vdottype, DOTPROD): New.
12218 (sur): Add SDOT and UDOT.
12220 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12222 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
12223 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
12224 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
12225 Add TARGET_DOTPROD.
12226 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
12227 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
12228 Enable TARGET_DOTPROD.
12229 (cortex-a75.cortex-a55): Likewise.
12230 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
12232 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12234 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
12235 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
12237 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
12239 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
12240 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
12241 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
12242 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
12243 * config/arm/types.md (neon_dot, neon_dot_q): New.
12244 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
12246 2017-10-16 Tamar Christina <tamar.christina@arm.com>
12248 * config/arm/arm.h (TARGET_DOTPROD): New.
12249 * config/arm/arm.c (arm_arch_dotprod): New.
12250 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
12251 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
12252 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
12253 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
12254 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
12255 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
12256 * doc/invoke.texi (armv8.2-a): Document dotprod
12258 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12260 * i386.c (ix86_vec_cost): New function.
12261 (ix86_rtx_costs): Handle vector operations better.
12262 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
12263 * x86-tune-costs.h: Add new costs to all tables.
12265 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
12267 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
12269 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
12270 divsd, sqrtss and sqrtsd
12271 * x86-tune-costs.h: Add new entries to all costs.
12272 (znver1_cost): Fix to match real instruction latencies.
12274 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12275 Michael Collison <michael.collison@arm.com>
12277 * compare-elim.c: Include emit-rtl.h.
12278 (can_merge_compare_into_arith): New function.
12279 (try_validate_parallel): Likewise.
12280 (try_merge_compare): Likewise.
12281 (try_eliminate_compare): Call the above when no previous clobber
12283 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
12286 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12288 PR middle-end/62263
12289 PR middle-end/82498
12290 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
12291 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
12293 PR middle-end/62263
12294 PR middle-end/82498
12295 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
12296 to be any operand_equal_p operands. For & (B - 1) require
12297 B to be power of 2. Recognize
12298 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
12300 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12303 * optabs.c (expand_memory_blockage): Fix call of
12304 targetm.have_memory_blockage.
12306 2017-10-14 Jakub Jelinek <jakub@redhat.com>
12309 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
12310 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
12311 objects to extra_objs instead of overwriting it.
12313 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12315 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
12316 Use any_fp_register_operand as operand[3] predicate. Simplify
12317 equality test for operands[2] and operands[4] memory location.
12318 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
12319 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
12320 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
12321 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
12322 any_fp_register_operand as operand[1] predicate. Simplify
12323 equality test for operands[0] and operands[3] memory location.
12324 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
12325 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
12326 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
12328 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
12330 * target-insns.def: Add memory_blockage.
12331 * optabs.c (expand_memory_blockage): New function.
12332 (expand_asm_memory_barrier): Rename ...
12333 (expand_asm_memory_blockage): ... to this.
12334 (expand_mem_thread_fence): Call expand_memory_blockage
12335 instead of expand_asm_memory_barrier.
12336 (expand_mem_singnal_fence): Ditto.
12337 (expand_atomic_load): Ditto.
12338 (expand_atomic_store): Ditto.
12339 * doc/md.texi (Standard Pattern Names For Generation):
12340 Document memory_blockage instruction pattern.
12342 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
12344 * config/rl78/rl78.c (rl78_emit_libcall): New function.
12345 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
12346 * config/rl78/rl78.md: New define_expand "adddi3".
12348 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12350 * cfghooks.c (verify_flow_info): Disable check that all probabilities
12353 2017-10-13 Jeff Law <law@redhat.com>
12355 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
12357 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12360 * internal-fn.c (expand_mul_overflow): If both operands have
12361 the same highpart of -1 or 0 and the topmost bit of lowpart
12362 is different, overflow is if res <= 0 rather than res < 0.
12364 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
12366 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
12367 TARGET_P9_VECTOR code for unaligned_load case.
12369 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
12371 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
12373 2017-10-13 Nathan Sidwell <nathan@acm.org>
12375 * tree-core.h (tree_contains_struct): Make bool.
12376 * tree.c (tree_contains_struct): Likewise.
12377 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
12378 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
12379 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
12380 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
12381 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
12383 2017-10-13 Richard Biener <rguenther@suse.de>
12385 * graphite-isl-ast-to-gimple.c
12386 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
12387 parameters and dominance check.
12388 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
12389 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
12390 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
12391 Do not update SSA form here or do intermediate IL verification.
12392 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
12393 (graphite_initialize): Remove check on the number of loops in
12394 the function and inline into graphite_transform_loops.
12395 (graphite_finalize): Inline into graphite_transform_loops.
12396 (graphite_transform_loops): Perform SSA update and IL verification
12398 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
12400 2017-10-13 Richard Biener <rguenther@suse.de>
12402 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
12403 graphite_expression_type_precision): Avoid global constructor
12405 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
12406 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
12407 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
12408 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
12409 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
12410 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
12412 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
12415 * config/i386/i386.h (ix86_red_zone_size): New.
12416 * config/i386/i386.md (push peephole2s): Replace
12417 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
12419 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12420 Alan Hayward <alan.hayward@arm.com>
12421 David Sherwood <david.sherwood@arm.com>
12423 * combine.c (can_change_dest_mode): Reject changes in
12424 REGMODE_NATURAL_SIZE.
12426 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12427 Alan Hayward <alan.hayward@arm.com>
12428 David Sherwood <david.sherwood@arm.com>
12430 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
12431 (expand_debug_source_expr): Likewise.
12432 * combine.c (combine_simplify_rtx): Likewise.
12433 * cse.c (fold_rtx): Likewise.
12434 * fwprop.c (canonicalize_address): Likewise.
12435 * targhooks.c (default_shift_truncation_mask): Likewise.
12437 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12438 Alan Hayward <alan.hayward@arm.com>
12439 David Sherwood <david.sherwood@arm.com>
12441 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
12442 (widened_mode): Likewise.
12443 (expand_unop): Likewise.
12444 * ree.c (transform_ifelse): Likewise.
12445 (merge_def_and_ext): Likewise.
12446 (combine_reaching_defs): Likewise.
12447 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12449 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12450 Alan Hayward <alan.hayward@arm.com>
12451 David Sherwood <david.sherwood@arm.com>
12453 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
12454 * combine.c (gen_lowpart_for_combine): Likewise.
12455 * dwarf2out.c (rtl_for_decl_location): Likewise.
12456 * final.c (alter_subreg): Likewise.
12457 * rtlhooks.c (gen_lowpart_general): Likewise.
12458 (gen_lowpart_if_possible): Likewise.
12460 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
12461 Alan Hayward <alan.hayward@arm.com>
12462 David Sherwood <david.sherwood@arm.com>
12464 * calls.c (expand_call): Use subreg_lowpart_offset.
12465 * cse.c (cse_insn): Likewise.
12466 * regcprop.c (copy_value): Likewise.
12467 (copyprop_hardreg_forward_1): Likewise.
12469 2017-10-13 Jakub Jelinek <jakub@redhat.com>
12472 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
12473 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
12474 =Q constraints to +Q and into insn condition add check
12475 that operands[0] and operands[1] are equal.
12476 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
12477 =Q constraints to +Q and into insn condition add check
12478 that operands[0] is equal to either operands[1] or operands[2].
12481 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
12482 instead of handling MINUS_EXPR twice (once for each argument),
12483 canonicalize operand order and handle just once, use rtype where
12484 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
12487 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
12488 any values of __C while still being pattern recognizable as a simple
12489 rotate instruction.
12491 2017-10-13 Richard Biener <rguenther@suse.de>
12493 PR tree-optimization/82451
12495 2017-10-02 Richard Biener <rguenther@suse.de>
12497 PR tree-optimization/82355
12498 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
12499 a mapping for the enclosing loop but avoid generating one for
12500 the loop tree root.
12501 (copy_bb_and_scalar_dependences): Remove premature codegen
12502 error on PHIs in blocks duplicated into multiple places.
12503 * graphite-scop-detection.c
12504 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
12505 in the region use it as loop and nest to analyze the DR in.
12506 (try_generate_gimple_bb): Likewise.
12507 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
12508 (add_loop_constraints): For blocks in a loop not in the region
12509 create a dimension with a single iteration.
12510 * sese.h (gbb_loop_at_index): Remove assert.
12512 * cfgloop.c (loop_preheader_edge): For the loop tree root
12513 return the single successor of the entry block.
12514 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
12515 Reset the SCEV hashtable and niters.
12516 * graphite-scop-detection.c
12517 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
12518 assert that we only have POLYNOMIAL_CHREC that vary in loops
12519 contained in the region.
12520 (scop_detection::graphite_can_represent_expr): Adjust.
12521 (scop_detection::stmt_has_simple_data_refs_p): For loops
12522 not in the region set loop to NULL. The nest is now the
12523 entry edge to the region.
12524 (try_generate_gimple_bb): Likewise.
12525 * sese.c (scalar_evolution_in_region): Adjust for
12526 instantiate_scev change.
12527 * tree-data-ref.h (graphite_find_data_references_in_stmt):
12528 Make nest parameter the edge into the region.
12529 (create_data_ref): Likewise.
12530 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
12531 entry edge into a region and adjust instantiate_scev calls.
12532 (create_data_ref): Likewise.
12533 (graphite_find_data_references_in_stmt): Likewise.
12534 (find_data_references_in_stmt): Pass the loop preheader edge
12535 from the nest argument.
12536 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
12537 parameter the edge into the region.
12538 (instantiate_parameters): Use the loop preheader edge as entry.
12539 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
12541 (get_instantiated_value_entry): Make instantiate_below parameter
12542 the edge into the region.
12543 (instantiate_scev_name): Likewise. Adjust dominance checks,
12544 when we cannot use loop-based instantiation instantiate by
12545 walking use-def chains.
12546 (instantiate_scev_poly): Adjust.
12547 (instantiate_scev_binary): Likewise.
12548 (instantiate_scev_convert): Likewise.
12549 (instantiate_scev_not): Likewise.
12550 (instantiate_array_ref): Remove.
12551 (instantiate_scev_3): Likewise.
12552 (instantiate_scev_2): Likewise.
12553 (instantiate_scev_1): Likewise.
12554 (instantiate_scev_r): Do not blindly handle N-operand trees.
12555 Do not instantiate array-refs. Handle all constants and invariants.
12556 (instantiate_scev): Make instantiate_below parameter
12557 the edge into the region.
12558 (resolve_mixers): Use the loop preheader edge for the region
12559 parameter to instantiate_scev_r.
12560 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
12562 2017-10-13 Richard Biener <rguenther@suse.de>
12564 PR tree-optimization/82525
12565 * graphite-isl-ast-to-gimple.c
12566 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
12568 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
12569 Fail code generation when we cannot represent the isl integer.
12570 (binary_op_to_tree): Elide modulo operations that are no-ops
12571 in the type we code generate. Remove now superfluous code
12574 2017-10-13 Richard Biener <rguenther@suse.de>
12576 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
12577 (scop_detection::harmful_loop_in_region): Remove premature
12578 IV type restriction.
12579 (scop_detection::graphite_can_represent_scev): We can handle
12580 pointer IVs just fine.
12582 2017-10-13 Alan Modra <amodra@gmail.com>
12584 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
12585 "Clobbers and Scratch Registers". Add paragraph on
12586 alternative to clobbers for scratch registers and OpenBLAS
12589 2017-10-13 Alan Modra <amodra@gmail.com>
12591 * doc/extend.texi (Clobbers): Correct vax example. Delete old
12592 example of a memory input for a string of known length. Move
12593 commentary out of table. Add a number of new examples
12594 covering array memory inputs.
12596 2017-10-12 Martin Liska <mliska@suse.cz>
12598 PR tree-optimization/82493
12599 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
12600 (test_range_functions): New function.
12601 (sbitmap_c_tests): Likewise.
12602 * selftest-run-tests.c (selftest::run_tests): Run new tests.
12603 * selftest.h (sbitmap_c_tests): New function.
12605 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
12607 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12609 * config/rs6000/amo.h: Fix spacing issue.
12611 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12614 * config/i386/i386.md (*ashl<mode>3_mask_1,
12615 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
12616 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
12619 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12621 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
12622 (profile_probability): Set max_probability
12623 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
12625 * profile-count.c (profile_probability::differs_from_p): Do not
12626 rely on max_probaiblity == 10000
12628 2017-10-12 Jeff Law <law@redhat.com>
12630 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
12633 2017-10-12 Martin Sebor <msebor@redhat.com>
12637 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
12638 (handle_alias_pairs): Call it.
12639 * common.opt (-Wattribute-alias): New option.
12640 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
12641 * doc/invoke.texi (-Wattribute-alias): Document.
12643 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
12646 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12648 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12650 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12651 (make_hard_regno_born, make_hard_regno_dead): Update
12652 bb_killed_pseudos and bb_gen_pseudos.
12654 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
12656 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
12658 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
12660 * config/alpha/alpha.c (alpha_split_conditional_move):
12661 Use std::swap instead of manually swapping.
12662 (alpha_stdarg_optimize_hook): Ditto.
12663 (alpha_canonicalize_comparison): Ditto.
12665 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12667 * tree-loop-distribution.c (struct builtin_info): New struct.
12668 (struct partition): Refactor fields into struct builtin_info.
12669 (partition_free): Free struct builtin_info.
12670 (build_size_arg_loc, build_addr_arg_loc): Delete.
12671 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
12672 information from struct builtin_info.
12673 (find_single_drs): New function refactored from classify_partition.
12674 Also moved builtin validity checks to this function.
12675 (compute_access_range, alloc_builtin): New functions.
12676 (classify_builtin_st, classify_builtin_ldst): New functions.
12677 (classify_partition): Refactor code into functions find_single_drs,
12678 classify_builtin_st and classify_builtin_ldst.
12679 (distribute_loop): Don't do runtime alias check when distributing
12681 (find_seed_stmts_for_distribution): New function.
12682 (pass_loop_distribution::execute): Refactor code finding seed
12683 stmts into above function. Support distribution for the innermost
12684 two-level loop nest. Adjust dump information.
12686 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12688 * tree-loop-distribution.c: Adjust the general comment.
12689 (NUM_PARTITION_THRESHOLD): New macro.
12690 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
12691 (classify_partition): Skip builtin pattern of loop nest's inner loop.
12692 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
12693 in call to build_partition_graph.
12694 (finalize_partitions): New parameter. Make loop distribution more
12695 conservative by fusing more partitions.
12696 (distribute_loop): Don't do runtime alias check in case of loop nest
12698 (find_seed_stmts_for_distribution): New function.
12699 (prepare_perfect_loop_nest): New function.
12700 (pass_loop_distribution::execute): Refactor code finding seed stmts
12701 and loop nest into above functions. Support loop nest distribution.
12702 Adjust dump information accordingly.
12704 2017-10-12 Bin Cheng <bin.cheng@arm.com>
12706 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
12707 and set PTYPE_SEQUENTIAL for merged partition.
12709 2017-10-12 Richard Biener <rguenther@suse.de>
12711 PR tree-optimization/69728
12713 2017-09-19 Richard Biener <rguenther@suse.de>
12715 PR tree-optimization/69728
12716 * graphite-sese-to-poly.c (schedule_error): New global.
12717 (add_loop_schedule): Handle empty domain by failing the
12719 (build_original_schedule): Handle schedule_error.
12721 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
12722 domain by returning an unchanged schedule.
12724 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12726 * genrecog.c (validate_pattern): For VEC_SELECT verify that
12727 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
12729 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
12731 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
12734 2017-10-12 Jakub Jelinek <jakub@redhat.com>
12737 * expr.c (store_field): Don't optimize away bitsize == 0 store
12738 from CALL_EXPR with addressable return type.
12740 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12742 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
12743 * config/rs6000/rs6000.md (sel): Delete mode attribute.
12744 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
12745 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
12746 TARGET_ISEL instead of TARGET_ISEL<sel>.
12748 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
12750 * config/rs6000/rs6000.c
12751 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
12753 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
12755 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
12757 (reg_or_cint_operand): Fix comment.
12758 (reg_or_zero_operand): New predicate.
12759 * config/rs6000/rs6000-protos.h (output_isel): Delete.
12760 * config/rs6000/rs6000.c (output_isel): Delete.
12761 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
12762 instead of reg_or_cint_operand. Output instruction directly (not via
12764 (isel_unsigned_<mode>): Ditto.
12765 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
12766 gpc_reg_operand. Add an instruction alternative for this. Output
12767 instruction directly.
12768 (*isel_reversed_unsigned_<mode>): Ditto.
12770 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12772 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
12773 (TARGET_CANONICALIZE_COMPARISON): Define.
12775 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
12778 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
12779 Check whether the dest is REG before adding REG_EQUIV note.
12781 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
12784 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
12786 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
12787 (make_hard_regno_born, make_hard_regno_dead): Update
12788 bb_killed_pseudos and bb_gen_pseudos.
12790 2017-10-11 Nathan Sidwell <nathan@acm.org>
12792 * incpath.h (enum incpath_kind): Name enum, prefix values.
12793 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
12794 * incpath.c (heads, tails): Use INC_MAX.
12795 (add_env_var_paths, add_standard_paths): Use incpath_kind.
12796 (merge_include_chains, split_quote_chain,
12797 register_include_chains): Update incpath_kind names.
12798 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
12799 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
12801 (add_framework_path, darwin_register_objc_includes): Likewise.
12802 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
12804 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12806 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
12807 Do not use float_operator operator predicate.
12808 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
12809 * config/i386/predicates.md (float_operator): Remove predicate.
12811 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
12813 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
12814 (*jccxf_i387): Ditto.
12815 (*jcc<mode>_i387): Ditto.
12816 (*jccu<mode>_i387): Ditto.
12817 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
12818 (*jcc_*_i387 splitters): Remove.
12819 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
12820 * config/i386/i386.c (ix86_split_fp_branch): Remove.
12821 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
12824 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12826 * profile-count.h (slow_safe_scale_64bit): New function.
12827 (safe_scale_64bit): New inline.
12828 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
12829 * profile-count.c: Include wide-int.h
12830 (slow_safe_scale_64bit): New.
12832 2017-10-11 Nathan Sidwell <nathan@acm.org>
12834 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
12835 HAS_DECL_ASSEMBLER_NAME_P.
12836 * gimple-expr.c (gimple_decl_printable_name: Check
12837 HAS_DECL_ASSEMBLER_NAME_P too.
12838 * ipa-utils.h (type_in_anonymous_namespace_p): Check
12839 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
12840 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
12841 * passes.c (rest_of_decl_compilation): Check
12842 HAS_DECL_ASSEMBLER_NAME_P too.
12843 * recog.c (verify_changes): Likewise.
12844 * tree-pretty-print.c (dump_decl_name): Likewise.
12845 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
12847 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
12848 (SET_DECL_ASSEMBLER_NAME): Use it.
12849 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
12850 (COPY_DECL_ASSEMBLER_NAME): Likewise.
12851 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
12853 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
12855 * config.gcc (i386, x86_64): Add extra objects.
12856 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
12857 (ix86_min_insn_size): Declare.
12858 (ix86_issue_rate): Declare.
12859 (ix86_adjust_cost): Declare.
12860 (ia32_multipass_dfa_lookahead): Declare.
12861 (ix86_macro_fusion_p): Declare.
12862 (ix86_macro_fusion_pair_p): Declare.
12863 (ix86_bd_has_dispatch): Declare.
12864 (ix86_bd_do_dispatch): Declare.
12865 (ix86_core2i7_init_hooks): Declare.
12866 (ix86_atom_sched_reorder): Declare.
12867 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
12868 (COSTS_N_BYTES): Move to x86-tune-costs.h.
12869 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
12870 (rip_relative_addr_p): Rename to ...
12871 (ix86_rip_relative_addr_p): ... this one; export.
12872 (memory_address_length): Update.
12873 (ix86_issue_rate): Move to x86-tune-sched.c.
12874 (ix86_flags_dependent): Move to x86-tune-sched.c.
12875 (ix86_agi_dependent): Move to x86-tune-sched.c.
12876 (exact_dependency_1): Move to x86-tune-sched.c.
12877 (exact_store_load_dependency): Move to x86-tune-sched.c.
12878 (ix86_adjust_cost): Move to x86-tune-sched.c.
12879 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
12880 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
12881 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
12882 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
12883 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
12884 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
12885 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
12886 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
12887 (min_insn_size): Rename to ...
12888 (ix86_min_insn_size): ... this one; export.
12889 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
12890 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
12891 (core2i7_first_cycle_multipass_backtrack): Move to
12892 x86-tune-sched-core.c.
12893 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
12894 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
12895 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
12896 (ix86_avoid_jump_mispredicts): Update.
12897 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
12898 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
12899 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
12900 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
12901 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
12902 (MAX_INSN): Move to ix86-tune-sched-bd.c.
12903 (MAX_IMM): Move to ix86-tune-sched-bd.c.
12904 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
12905 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
12906 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
12907 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
12908 (MAX_STORE): Move to ix86-tune-sched-bd.c.
12909 (BIG): Move to ix86-tune-sched-bd.c.
12910 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
12911 (enum insn_path): Move to ix86-tune-sched-bd.c.
12912 (get_mem_group): Move to ix86-tune-sched-bd.c.
12913 (is_cmp): Move to ix86-tune-sched-bd.c.
12914 (dispatch_violation): Move to ix86-tune-sched-bd.c.
12915 (is_branch): Move to ix86-tune-sched-bd.c.
12916 (is_prefetch): Move to ix86-tune-sched-bd.c.
12917 (init_window): Move to ix86-tune-sched-bd.c.
12918 (allocate_window): Move to ix86-tune-sched-bd.c.
12919 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
12920 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
12921 (process_end_window): Move to ix86-tune-sched-bd.c.
12922 (allocate_next_window): Move to ix86-tune-sched-bd.c.
12923 (find_constant): Move to ix86-tune-sched-bd.c.
12924 (get_num_immediates): Move to ix86-tune-sched-bd.c.
12925 (has_immediate): Move to ix86-tune-sched-bd.c.
12926 (get_insn_path): Move to ix86-tune-sched-bd.c.
12927 (get_insn_group): Move to ix86-tune-sched-bd.c.
12928 (count_num_restricted): Move to ix86-tune-sched-bd.c.
12929 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
12930 (add_insn_window): Move to ix86-tune-sched-bd.c.
12931 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
12932 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
12933 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
12934 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
12935 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
12936 (do_dispatch): Move to ix86-tune-sched-bd.c.
12937 (has_dispatch): Move to ix86-tune-sched-bd.c.
12938 * config/i386/t-i386: Add new object files.
12939 * config/i386/x86-tune-costs.h: New file.
12940 * config/i386/x86-tune-sched-atom.c: New file.
12941 * config/i386/x86-tune-sched-bd.c: New file.
12942 * config/i386/x86-tune-sched-core.c: New file.
12943 * config/i386/x86-tune-sched.c: New file.
12945 2017-10-11 Liu Hao <lh_mouse@126.com>
12947 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
12948 the generic version below instead.
12949 (should_colorize): Recognize Windows consoles as terminals
12951 * pretty-print.c [__MINGW32__] (write_all): New function.
12952 [__MINGW32__] (find_esc_head): Likewise.
12953 [__MINGW32__] (find_esc_terminator): Likewise.
12954 [__MINGW32__] (eat_esc_sequence): Likewise.
12955 [__MINGW32__] (mingw_ansi_fputs): New function that handles
12957 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
12960 2017-10-11 Richard Biener <rguenther@suse.de>
12962 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
12963 Properly call analyze_scalar_evolution with the loop of the stmt.
12965 2017-10-11 Richard Biener <rguenther@suse.de>
12967 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
12968 * tree-core.h (tree_base): Add chrec_var union member.
12969 * tree.h (CHREC_VAR): Remove.
12970 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
12971 * tree-chrec.h (build_polynomial_chrec): Adjust.
12972 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
12973 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
12975 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
12977 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
12978 * match.pd: ... here.
12979 ((T) X == (T) Y): Relax condition.
12981 2017-10-11 Bin Cheng <bin.cheng@arm.com>
12983 PR tree-optimization/82472
12984 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
12986 (break_alias_scc_partitions): Update postorder number.
12988 2017-10-11 Martin Liska <mliska@suse.cz>
12991 * opts.c (parse_no_sanitize_attribute): Do not use error_value
12993 * opts.h (parse_no_sanitize_attribute): Remove last argument.
12995 2017-10-11 Martin Liska <mliska@suse.cz>
12997 * print-rtl.c (print_insn): Move declaration of idbuf
12998 to same scope as name.
13000 2017-10-11 Martin Liska <mliska@suse.cz>
13005 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
13006 volatile arguments.
13008 2017-10-11 Martin Liska <mliska@suse.cz>
13011 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
13012 volatile arguments.
13014 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
13016 * config.gcc (default_gnu_indirect_function): Default to yes for
13017 arm*-*-linux* with glibc.
13019 2017-10-11 Richard Biener <rguenther@suse.de>
13021 * tree-scalar-evolution.c (get_scalar_evolution): Handle
13022 default-defs and types we do not want to analyze.
13023 (interpret_loop_phi): Replace unreachable code with an assert.
13024 (compute_scalar_evolution_in_loop): Remove and inline ...
13025 (analyze_scalar_evolution_1): ... here, replacing condition with
13026 what makes the intent clearer. Remove handling of cases
13027 get_scalar_evolution now handles.
13029 2017-10-10 Jim Wilson <wilson@tuliptree.org>
13031 PR rtl-optimization/81434
13032 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
13033 comment for main loop. In sched_group_found if, also add checks for
13034 pass and min_cost_group.
13036 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
13038 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
13039 (rs6000_insn_cost): New function.
13040 * config/rs6000/rs6000.md (cost): New attribute.
13042 2017-10-10 Jakub Jelinek <jakub@redhat.com>
13043 H.J. Lu <hongjiu.lu@intel.com>
13047 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
13048 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
13049 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
13050 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
13051 to OPTION_MASK_ISA_AVX512VL - builtins that have both
13052 OPTION_MASK_ISA_MMX and some other bit set require both
13053 mmx and the ISAs without the mmx bit.
13054 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
13055 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
13056 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
13057 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
13058 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
13059 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
13060 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
13061 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
13062 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
13063 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
13064 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
13065 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
13066 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
13067 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
13068 Add OPTION_MASK_ISA_MMX.
13070 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
13072 * config.gcc (armv7*-*-freebsd*): New target.
13073 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
13075 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
13077 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
13078 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
13081 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
13083 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
13084 a template parameter.
13085 (WIDE_INT_REF_FOR): Update accordingly.
13086 * tree.h (wi::int_traits <const_tree>): Delete.
13087 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
13088 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
13089 (wi::tree_to_wide_ref): New typedef.
13090 (wi::to_wide): New function.
13091 * calls.c (get_size_range): Use wi::to_wide when operating on
13092 trees as wide_ints.
13093 * cgraph.c (cgraph_node::create_thunk): Likewise.
13094 * config/i386/i386.c (ix86_data_alignment): Likewise.
13095 (ix86_local_alignment): Likewise.
13096 * dbxout.c (stabstr_O): Likewise.
13097 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
13098 * expr.c (const_vector_from_tree): Likewise.
13099 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
13100 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
13101 (fold_negate_expr_1, int_const_binop_1, const_binop)
13102 (fold_convert_const_int_from_real, optimize_bit_field_compare)
13103 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
13104 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
13105 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
13106 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
13107 (fold_not_const, round_up_loc): Likewise.
13108 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
13109 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
13110 (alloca_call_type): Likewise.
13111 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
13112 * godump.c (go_output_typedef): Likewise.
13113 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
13114 * internal-fn.c (get_min_precision): Likewise.
13115 * ipa-cp.c (ipcp_store_vr_results): Likewise.
13116 * ipa-polymorphic-call.c
13117 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
13118 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
13119 (ipa_modify_call_arguments): Likewise.
13120 * match.pd: Likewise.
13121 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
13122 * print-tree.c (print_node_brief, print_node): Likewise.
13123 * stmt.c (expand_case): Likewise.
13124 * stor-layout.c (layout_type): Likewise.
13125 * tree-affine.c (tree_to_aff_combination): Likewise.
13126 * tree-cfg.c (group_case_labels_stmt): Likewise.
13127 * tree-data-ref.c (dr_analyze_indices): Likewise.
13128 (prune_runtime_alias_test_list): Likewise.
13129 * tree-dump.c (dequeue_and_dump): Likewise.
13130 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
13131 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
13132 * tree-pretty-print.c (dump_generic_node): Likewise.
13133 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
13134 (simple_iv_with_niters): Likewise.
13135 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
13136 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
13137 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
13138 * tree-ssa-loop-niter.c (split_to_var_and_offset)
13139 (refine_value_range_using_guard, number_of_iterations_ne_max)
13140 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
13141 (get_cst_init_from_scev, record_nonwrapping_iv)
13142 (scev_var_range_cant_overflow): Likewise.
13143 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
13144 * tree-ssa-pre.c (compute_avail): Likewise.
13145 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
13146 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
13147 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
13148 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
13149 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
13150 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
13151 (dump_case_nodes, try_switch_expansion): Likewise.
13152 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
13153 (vect_do_peeling): Likewise.
13154 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
13155 * tree-vect-stmts.c (vectorizable_load): Likewise.
13156 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
13157 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
13158 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
13159 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
13160 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
13161 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
13162 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
13163 (evrp_dom_walker::before_dom_children): Likewise.
13164 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
13165 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
13166 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
13167 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
13168 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
13169 (get_range_pos_neg): Likewise.
13170 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
13171 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
13172 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
13173 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
13174 * config/avr/avr.c (avr_fold_builtin): Likewise.
13175 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
13176 * config/msp430/msp430.c (msp430_attr): Likewise.
13177 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
13178 * config/powerpcspe/powerpcspe-c.c
13179 (altivec_resolve_overloaded_builtin): Likewise.
13180 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
13181 (rs6000_expand_ternop_builtin): Likewise.
13182 * config/rs6000/rs6000-c.c
13183 (altivec_resolve_overloaded_builtin): Likewise.
13184 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
13185 (rs6000_expand_ternop_builtin): Likewise.
13186 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
13188 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13190 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
13191 when copying loop nest with only one inner loop.
13193 2017-10-10 Richard Biener <rguenther@suse.de>
13195 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
13196 blocks if SCEV is active.
13197 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
13199 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
13200 (scev_initialize): Assert we are not yet initialized.
13202 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13204 * tree-loop-distribution.c (generate_loops_for_partition): Remove
13205 inner loop's exit stmt by making it always exit the loop, otherwise
13206 we would generate an infinite empty loop.
13208 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13210 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
13211 renaming variables in new preheader if it's deleted.
13213 2017-10-10 Bin Cheng <bin.cheng@arm.com>
13215 * tree-loop-distribution.c (struct partition): Remove unused field
13216 loops of the structure.
13217 (partition_alloc, partition_free): Ditto.
13218 (build_rdg_partition_for_vertex): Ditto.
13220 2017-10-09 Jeff Law <law@redhat.com>
13222 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
13223 return type to match prototype and documentation.
13225 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13227 * config/rs6000/rs6000.c (processor_costs): Move to ...
13228 * config/rs6000/rs6000.h: ... here.
13229 (rs6000_cost): Declare.
13231 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
13233 * except.c (setjmp_fn): New global variable.
13234 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
13235 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
13236 if DONT_USE_BUILTIN_SETJMP is defined.
13238 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13240 * target.def (insn_cost): New hook.
13241 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
13242 * doc/tm.texi: Regenerate.
13243 * rtlanal.c (insn_cost): Use the new hook.
13245 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13247 * combine.c (combine_validate_cost): Compute the new insn_cost,
13248 not just pattern_cost.
13249 (try_combine): Adjust comment.
13251 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
13253 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
13255 * combine.c (uid_insn_cost): Adjust comment.
13256 (combine_validate_cost): Adjust comment. Use pattern_cost instead
13258 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
13259 * dse.c (find_shift_sequence): Ditto.
13260 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
13261 (bb_valid_for_noce_process_p): Use pattern_cost.
13262 * rtl.h (insn_rtx_cost): Delete.
13263 (pattern_cost): New prototype.
13264 (insn_cost): New prototype.
13265 * rtlanal.c (insn_rtx_cost): Rename to...
13266 (pattern_cost): ... this.
13269 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
13271 * config/i386/i386.md (*jcc_2): Remove insn pattern.
13272 (*jcc<mode>_0_r_i387): Ditto.
13273 (*jccxf_r_i387): Ditto.
13274 (*jcc<mode>_r_i387): Ditto.
13275 (*jccu<mode>_r_i387): Ditto.
13276 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
13277 (*jcc): Rename from *jcc_1.
13279 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13281 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
13282 deferred rescans after the lvx/stvx recombination pre-pass.
13284 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
13286 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
13287 memory operation instruction support.
13288 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
13289 (rs6000-ibm-aix[789]*): Likewise.
13290 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
13291 Document new functions.
13293 2017-10-09 Richard Biener <rguenther@suse.de>
13295 PR tree-optimization/82397
13296 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
13297 equality only for semantically equal trees.
13299 2017-10-09 Richard Biener <rguenther@suse.de>
13301 PR tree-optimization/82449
13302 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
13303 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
13304 allow constant addresses.
13305 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
13308 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13310 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
13313 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13316 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
13319 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13322 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
13324 2017-10-09 Jakub Jelinek <jakub@redhat.com>
13327 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
13328 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
13330 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
13332 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
13333 (WI_BINARY_PREDICATE_RESULT): Likewise.
13334 (wi::binary_traits::operator_result): New type.
13335 (wi::binary_traits::predicate_result): Likewise.
13336 (generic_wide_int::operator~, unary generic_wide_int::operator-)
13337 (generic_wide_int::operator==, generic_wide_int::operator!=)
13338 (generic_wide_int::operator&, generic_wide_int::and_not)
13339 (generic_wide_int::operator|, generic_wide_int::or_not)
13340 (generic_wide_int::operator^, generic_wide_int::operator+
13341 (binary generic_wide_int::operator-, generic_wide_int::operator*):
13343 (operator~, unary operator-, operator==, operator!=, operator&)
13344 (operator|, operator^, operator+, binary operator-, operator*): New
13346 * expr.c (get_inner_reference): Use wi::bit_and_not.
13347 * fold-const.c (fold_binary_loc): Likewise.
13348 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
13349 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
13350 (bit_value_binop): Likewise.
13351 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
13352 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
13353 (extract_range_from_binary_expr_1): Likewise.
13354 (masked_increment): Likewise.
13355 (simplify_bit_ops_using_ranges): Likewise.
13357 2017-10-09 Martin Jambor <mjambor@suse.cz>
13360 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
13361 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
13362 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
13364 (hsa_fixup_mov_insn_type): New function.
13365 (hsa_op_with_type::get_in_type): Use it.
13366 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
13367 immediates in an assert.
13368 (hsa_op_with_type::extend_int_to_32bit): New method.
13369 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
13370 types. Convert to dest type if necessary.
13371 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
13372 (reg_for_gimple_ssa): Pass false as min32int to
13373 hsa_type_for_scalar_tree_type.
13374 (gen_hsa_addr): Fixup type when creating addresable temporary.
13375 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
13376 (gen_hsa_unary_operation): Extend operands and convert to dest type if
13377 necessary. Call hsa_fixup_mov_insn_type.
13378 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
13379 extend operands and convert to dest type if necessary.
13380 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
13381 to dest type if necessary.
13382 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
13383 if conversion nt necessary and size matches.
13384 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
13385 to dest type if necessary.
13386 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
13387 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
13389 (gen_hsa_clrsb): Likewise.
13390 (gen_hsa_ffs): Likewise.
13391 (gen_hsa_divmod): Extend operands and convert to dest type if
13393 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
13395 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13397 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
13398 Remove empty default arguments. Use a brace block as output
13400 (conditional return): Ditto.
13402 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
13403 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
13405 (group_ending_nop): Ditto.
13406 (doloop_end): Ditto.
13407 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
13408 (splitters for those): Ditto.
13410 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
13412 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
13413 a conditional jump (and the compare for it) so that pc_rtx is the
13415 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
13416 for the deleted and renamed ctr<mode>_internal[234] patterns.
13417 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
13418 Delete second conditional return pattern.
13419 (ctr<mode>_internal2): Delete this second bdnz pattern.
13420 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
13421 (ctr<mode>_internal4): Delete this second bdz pattern.
13423 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13425 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
13426 (always_initialized_rtx_for_ssa_name_p): New predicate.
13427 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
13428 (finish_out_of_ssa): Free new field of SA.
13429 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
13430 * tree-ssa-coalesce.c: Include tree-ssa.h.
13431 (get_parm_default_def_partitions): Remove extern keyword.
13432 (get_undefined_value_partitions): New function.
13433 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
13434 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
13435 uninitialized bits.
13436 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
13438 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
13440 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
13442 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13444 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
13445 for targets that preffer 128bit.
13447 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
13449 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
13451 2017-10-08 Olivier Hainque <hainque@adacore.com>
13453 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
13454 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
13456 (thumb_set_return_address): Likewise.
13458 2017-10-08 Olivier Hainque <hainque@adacore.com>
13460 * common/config/arm/arm-common.c (arm_except_unwind_info):
13461 Handle DWARF2_UNWIND_INFO.
13463 2017-10-07 Michael Collison <michael.collison@arm.com>
13465 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
13468 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
13470 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
13471 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
13472 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
13473 defined, force the creation of a new block for a dispatch label.
13475 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
13477 * invoke.texi (Wsuggest-attribute=cold): Document.
13478 * common.opt (Wsuggest-attribute=cold): New
13479 * ipa-pure-const.c (warn_function_cold): New function.
13480 * predict.c (compute_function_frequency): Use it.
13481 * predict.h (warn_function_cold): Declare.
13483 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
13485 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
13488 2017-10-06 Martin Liska <mliska@suse.cz>
13490 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
13491 keyword for member functions.
13492 (struct sanopt_tree_couple): New struct.
13493 (struct sanopt_tree_couple_hash): New function.
13494 (struct sanopt_ctx): Add new hash_map.
13495 (has_dominating_ubsan_ptr_check): New function.
13496 (record_ubsan_ptr_check_stmt): Likewise.
13497 (maybe_optimize_ubsan_ptr_ifn): Likewise.
13498 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
13499 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
13501 2017-10-06 Sudakshina Das <sudi.das@arm.com>
13504 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
13505 aarch64_simd_valid_immediate on CONST_VECTORs.
13506 (aarch64_reg_or_bic_imm): Likewise.
13508 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
13510 PR rtl-optimization/82396
13511 * haifa-sched.c (ready_sort_real): Disable qsort checking.
13513 2017-10-06 Sebastian Pop <sebpop@gmail.com>
13515 * graphite-dependences.c (scop_get_reads): Move code to...
13516 (scop_get_must_writes): Move code to...
13517 (scop_get_may_writes): Move code to...
13518 (scop_get_reads_and_writes): ... here.
13519 (scop_get_dependences): Call scop_get_reads_and_writes.
13521 2017-10-06 Jakub Jelinek <jakub@redhat.com>
13523 PR tree-optimization/82434
13524 * fold-const.h (can_native_encode_type_p,
13525 can_native_encode_string_p): Remove.
13526 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
13527 don't encode anything, just return what would be otherwise returned.
13528 (native_encode_fixed, native_encode_complex, native_encode_vector):
13530 (native_encode_string): Likewise. Inline by hand
13531 can_native_encode_string_p.
13532 (can_native_encode_type_p): Remove.
13533 (can_native_encode_string_p): Remove.
13534 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
13535 STRING_CSTs using can_native_encode_string_p, test all
13536 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
13537 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
13538 argument from native_encode_expr.
13539 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
13540 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
13543 2017-10-06 Richard Biener <rguenther@suse.de>
13545 PR tree-optimization/82397
13546 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
13547 operand_equal_p but rely on data_ref_compare_tree for detecting
13549 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
13550 to match up with dr_group_sort_cmp.
13552 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13555 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
13557 * config/s390/s390-builtin-types.def: Regenerate.
13559 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13562 * config/s390/s390-builtin-types.def: Regenerate.
13563 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
13564 Change flag from B_VXE to B_VX.
13565 (s390_vec_min_dbl): Remove B_VXE flag.
13567 2017-10-06 Richard Biener <rguenther@suse.de>
13569 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
13570 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
13571 translate_isl_ast_to_gimple::is_valid_rename,
13572 translate_isl_ast_to_gimple::get_rename,
13573 translate_isl_ast_to_gimple::get_def_bb_for_const,
13574 translate_isl_ast_to_gimple::get_new_name,
13575 translate_isl_ast_to_gimple::collect_all_ssa_names,
13576 translate_isl_ast_to_gimple::copy_loop_phi_args,
13577 translate_isl_ast_to_gimple::collect_all_ssa_names,
13578 translate_isl_ast_to_gimple::copy_loop_phi_args,
13579 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
13580 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
13581 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
13582 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
13583 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
13584 translate_isl_ast_to_gimple::copy_cond_phi_args,
13585 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
13586 translate_isl_ast_to_gimple::edge_for_new_close_phis,
13587 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
13588 translate_isl_ast_to_gimple::rename_uses,
13589 translate_isl_ast_to_gimple::rename_all_uses): Remove.
13590 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
13591 (set_rename_for_each_def): Likewise.
13592 (graphite_copy_stmts_from_block): Handle debug stmt resetting
13593 here. Handle rewriting SCEV analyzable uses here.
13594 (copy_bb_and_scalar_dependences): Generate code for PHI
13596 (graphite_regenerate_ast_isl): Adjust.
13597 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
13598 (add_write, add_read): New functions.
13599 (build_cross_bb_scalars_def): Use it and simplify.
13600 (build_cross_bb_scalars_use): Likewise.
13601 (graphite_find_cross_bb_scalar_vars): Inline into...
13602 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
13603 simulating out-of-SSA. Compute liveout and add dependencies.
13604 (build_scops): Force an empty entry block.
13605 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
13607 (sese_build_liveouts): Declare.
13608 (sese_trivially_empty_bb_p): Likewise.
13609 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
13610 compute liveout and debug_liveout.
13611 (sese_bad_liveouts_use): Remove.
13612 (sese_reset_debug_liveouts_bb): Likewise.
13613 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
13614 (sese_build_liveouts): Build liveout and debug_liveout and store
13616 (new_sese_info): Adjust.
13617 (free_sese_info): Likewise.
13618 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
13619 do not build liveout here.
13620 (move_sese_in_condition): Adjust region entry.
13621 (scev_analyzable_p): Match up with chrec_apply requirements.
13622 (sese_trivially_empty_bb_p): New.
13623 * tree-into-ssa.c (get_reaching_def): Properly support generating
13624 default-defs for incremental rewrite of anonymous names.
13626 2017-10-06 Richard Biener <rguenther@suse.de>
13628 * graphite-sese-to-poly.c (extract_affine): For casts increasing
13629 precision do not perform modulo reduction.
13631 2017-10-06 Richard Biener <rguenther@suse.de>
13633 PR tree-optimization/82436
13634 * tree-vect-slp.c (vect_supported_load_permutation_p): More
13635 conservatively choose the vectorization factor when checking
13636 whether we can perform the required load permutation.
13637 (vect_transform_slp_perm_load): Assert when we may not fail.
13639 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
13641 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
13642 message for incompatible -msdata=* and -mcall-* options.
13644 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13646 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
13647 rate for post-reload scheduling.
13649 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13651 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
13653 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13655 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
13656 to improve monte carlo in scimark.
13658 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
13660 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
13661 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
13662 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
13663 pentium4_cost, nocona_cost): Set reassociation width to 1.
13664 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
13665 width to 2 for fp operations and 1 otherwise.
13666 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
13668 (atom_cost): Set reassociation width to 2.
13669 (slm_cost, generic_cost): Set fp reassociation width
13670 to 2 and 1 otherwise.
13671 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
13672 (core_cost): Set fp reassociation width to 4 and vector to 2.
13673 (ix86_reassociation_width): Rewrite using cost table; special case
13674 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
13675 and TARGET_AVX128_OPTIMAL.
13676 * config/i386/i386.h (processor_costs): Add
13677 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
13678 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
13679 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
13680 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
13681 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
13682 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
13684 2017-10-05 Nathan Sidwell <nathan@acm.org>
13686 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
13688 2017-10-05 Tamar Christina <tamar.christina@arm.com>
13690 * config/arm/arm.c (arm_test_fpu_data): New.
13691 (arm_run_selftests): Call arm_test_fpu_data.
13693 2017-10-04 Nathan Sidwell <nathan@acm.org>
13695 * toplev.c (toplev::main): Remove excess parens on pretty_printer
13697 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
13699 2017-10-04 Sudakshina Das <sudi.das@arm.com>
13701 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
13702 check type for aarch64_simd_valid_immediate.
13703 (aarch64_output_simd_mov_immediate): Update prototype.
13704 (aarch64_simd_valid_immediate): Update prototype.
13705 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
13706 support for ORR-immediate.
13707 (and<mode>3): modified pattern to add support for BIC-immediate.
13708 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
13709 now checks for valid immediate for BIC and ORR based on new enum
13711 (aarch64_output_simd_mov_immediate): Function now used to output
13712 BIC/ORR imm as well based on new enum argument.
13713 * config/aarch64/constraints.md (Do): New vector immediate constraint.
13715 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
13716 (aarch64_reg_or_bic_imm): Likewise.
13718 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13720 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
13721 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
13723 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13727 PR rtl-optimization/82396
13728 * haifa-sched.c (autopref_multipass_init): Simplify
13730 (autopref_rank_data): Simplify sort order.
13731 * sched-int.h (autopref_multipass_data_): Remove
13732 multi_mem_insn_p, min_offset and max_offset.
13734 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13736 * doc/sourcebuild.texi: Document vect_peeling_profitable.
13738 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13740 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
13741 vect_doubleint_cvt.
13743 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13745 * doc/sourcebuild.texi: Document vect_long_mult.
13747 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
13749 PR tree-optimization/82413
13750 * fold-const.c (build_range_check): Use widest_int when comparing
13751 the maximum ETYPE value with HIGH.
13753 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
13755 PR rtl-optimization/82396
13756 * haifa-sched.c (autopref_multipass_init): Simplify
13758 (autopref_rank_data): Simplify sort order.
13759 * sched-int.h (autopref_multipass_data_): Remove
13760 multi_mem_insn_p, min_offset and max_offset.
13762 2017-10-04 Jakub Jelinek <jakub@redhat.com>
13764 PR tree-optimization/82381
13765 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
13766 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
13769 PR tree-optimization/82374
13770 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
13771 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
13772 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
13773 current_function_decl to the new decl.
13775 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
13777 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
13778 helper macro for IEEE float128 hardware built-in functions.
13779 (SQRTF128_ODD): Add built-in functions with the round-to-odd
13781 (TRUNCF128_ODD): Likewise.
13782 (ADDF128_ODD): Likewise.
13783 (SUBF128_ODD): Likewise.
13784 (MULF128_ODD): Likewise.
13785 (DIVF128_ODD): Likewise.
13786 (FMAF128_ODD): Likewise.
13787 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
13788 UNSPEC_TRUNC_ROUND_TO_ODD.
13789 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
13790 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
13791 floating point round to odd instructions.
13792 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
13793 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
13794 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
13795 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
13796 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
13797 (trunc<mode>sf2_hw): Change the truncate with round to odd
13798 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
13799 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
13800 to odd hardware instructions.
13801 (sub<mode>3_odd): Likewise.
13802 (mul<mode>3_odd): Likewise.
13803 (div<mode>3_odd): Likewise.
13804 (sqrt<mode>2_odd): Likewise.
13805 (fma<mode>4_odd): Likewise.
13806 (fms<mode>4_odd): Likewise.
13807 (nfma<mode>4_odd): Likewise.
13808 (nfms<mode>4_odd): Likewise.
13809 (trunc<mode>df2_odd): Change the truncate with round to odd
13810 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
13812 * doc/extend.texi (PowerPC built-in functions): Update documentation
13813 for existing IEEE float128-bit built-in functions. Add built-in
13814 functions that generate the IEEE 128-bit floating point round to
13817 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
13819 PR rtl-optimization/77729
13820 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
13821 to (X&(C1&~C2))|C2 transformations.
13823 2017-10-03 Martin Jambor <mjambor@suse.cz>
13825 PR tree-optimization/82363
13826 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
13827 mismatch, mark lacc written regardless of racc.
13829 2017-10-03 Jakub Jelinek <jakub@redhat.com>
13831 PR tree-optimization/82381
13832 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
13833 stmt_to_insert nor wheather SSA_NAMEs are default defs.
13834 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
13835 fallthrough into reassoc_stmt_dominates_stmt_p.
13838 * combine.c (combine_instructions): Don't combine in unreachable
13841 2017-10-02 Peter Bergner <bergner@vnet.ibm.com>
13844 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
13845 function to not use the have_cpu variable. Do not set cpu_index,
13846 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
13847 or the default cpu.
13848 (rs6000_valid_attribute_p): Remove duplicate initializations of
13849 old_optimize and func_optimize.
13850 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
13851 (rs6000_activate_target_options): Make global.
13852 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
13855 2017-10-02 Jakub Jelinek <jakub@redhat.com>
13857 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
13858 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
13859 Set *poffset to 0 and *psize and *pmax_size to -1 if
13860 *poffset + *psize overflows in HOST_WIDE_INT.
13862 PR tree-optimization/82387
13863 PR tree-optimization/82388
13864 PR tree-optimization/82389
13865 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
13866 instead of live_bytes non-NULL.
13868 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
13871 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
13874 2017-10-02 Richard Biener <rguenther@suse.de>
13876 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
13877 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
13878 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
13880 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13882 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
13883 requested precision matches the type's.
13884 * calls.c (alloc_max_size): Calculate the new candidate size as
13885 a widest_int and use wi::to_widest when comparing it with the
13886 current candidate size.
13887 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
13888 zero rather than integer_zero_node.
13889 * match.pd: Check for a no-op conversion before using wi::add
13890 rather than after. Use tree_to_uhwi when summing small shift
13891 counts into an unsigned int.
13893 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
13894 Alan Hayward <alan.hayward@arm.com>
13895 David Sherwood <david.sherwood@arm.com>
13898 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
13899 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
13900 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
13901 POINTER_AND_FP_REGS.
13903 2017-10-02 Richard Biener <rguenther@suse.de>
13905 PR tree-optimization/82355
13906 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
13907 a mapping for the enclosing loop but avoid generating one for
13908 the loop tree root.
13909 (copy_bb_and_scalar_dependences): Remove premature codegen
13910 error on PHIs in blocks duplicated into multiple places.
13911 * graphite-scop-detection.c
13912 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
13913 in the region use it as loop and nest to analyze the DR in.
13914 (try_generate_gimple_bb): Likewise.
13915 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
13916 (add_loop_constraints): For blocks in a loop not in the region
13917 create a dimension with a single iteration.
13918 * sese.h (gbb_loop_at_index): Remove assert.
13920 2017-10-01 Kevin Buettner <kevinb@redhat.com>
13922 * omp-expand.c (adjust_context_scope): New function.
13923 (expand_parallel_call): Call adjust_context_scope.
13925 2017-10-01 Jeff Law <law@redhat.com>
13927 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
13928 dom_opt_dom_walker class with direct access to private members.
13929 Add comments. Call test_for_singularity.
13930 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
13931 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
13932 m_dummy_cond anymore.
13933 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
13935 (pass_dominator:execute): Build the dummy_cond here and pass it
13936 to the dom_opt_dom_walker ctor.
13937 (test_for_singularity): New function.
13939 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
13940 Maya Rashish <coypu@sdf.org>
13942 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
13943 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
13944 (alpha*-*-netbsd*) Use nbsd_tm_file.
13945 (arm*-*-netbsdelf*) Likewise.
13946 (i[34567]86-*-netbsdelf*) Likewise.
13947 (x86_64-*-netbsd*) Likewise.
13948 (mips*-*-netbsd*) Likewise.
13949 (powerpc-*-netbsd*) Likewise.
13950 (sh*-*-netbsd*) Likewise.
13951 (sparc-*-netbsdelf*) Likewise.
13952 (sparc64-*-netbsd*) Likewise.
13953 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
13955 (vax-*-netbsdelf*) Likewise.
13956 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
13957 (UINT_FAST8_TYPE) Likewise.
13958 (INT_FAST16_TYPE) Check CHAR_FAST16.
13959 (UINT_FAST16_TYPE) Likewise.
13961 2017-09-30 Jakub Jelinek <jakub@redhat.com>
13964 * config/i386/i386.md
13965 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
13966 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
13967 *divmodsi4_zext_2): New define_insn_and_split.
13968 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
13969 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
13970 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
13971 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
13972 New define_insn_and_split.
13973 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
13974 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
13975 operands[1] having DImode when mode is SImode.
13977 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
13978 always SImode for DIV and MOD in REG_EQUAL notes.
13980 2017-09-29 Yury Gribov <tetra2005@gmail.com>
13982 PR middle-end/82319
13983 * match.pd: Fix handling of NaNs in pattern.
13985 2017-09-29 Jeff Law <law@redhat.com>
13987 * sbitmap.c (bitmap_bit_in_range_p): New function.
13988 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
13989 * tree-ssa-dse.c (live_bytes_read): New function.
13990 (dse_classify_store): Ignore reads of dead bytes.
13992 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
13993 typos and whitespace errors.
13994 * config/i386/predicates.md (address_no_seg_operand): Likewise.
13995 * config/s390/s390.c (s390_emit_prologue): Likewise.
13997 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
14000 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
14001 with a symbol for LRA.
14003 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
14005 PR rtl-optimization/82338
14006 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
14008 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
14010 * genmodes.c (calc_wider_mode): Suppress qsort macro.
14011 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
14012 (qsort_chk): Declare.
14013 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
14014 (qsort_chk): New function.
14016 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14018 PR tree-optimization/82337
14019 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
14020 phi definition if the PHI result appears in an abnormal PHI.
14021 (find_basis_for_base_expr): Don't record a basis if the LHS of the
14022 basis appears in an abnormal PHI.
14024 2017-09-29 Richard Biener <rguenther@suse.de>
14026 * graphite-isl-ast-to-gimple.c
14027 (translate_isl_ast_to_gimple::set_codegen_error): New function.
14028 (binary_op_to_tree): Use it.
14029 (get_rename_from_scev): Likewise.
14030 (copy_loop_phi_nodes): Likewise.
14031 (copy_bb_and_scalar_dependences): Likewise.
14032 (translate_pending_phi_nodes): Likewise.
14034 2017-09-29 Jakub Jelinek <jakub@redhat.com>
14037 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
14038 for movabsq $(i32 << shift), r64.
14040 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
14042 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
14043 index when encoding %esp as %rsp to avoid 0x67 prefix.
14045 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
14047 * config/i386/i386.md (*movsf_internal, *movdf_internal):
14048 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
14050 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
14052 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
14053 Extensions with more than 16 double VFP registers.
14054 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
14055 to_clear_mask and all code related to it. Replace the remaining
14056 entry by a sbitmap and adapt code accordingly.
14058 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
14060 * brig-builtins.def: Change pure attributes to const.
14062 2017-09-28 Joseph Myers <joseph@codesourcery.com>
14064 * config.gcc (default_gnu_indirect_function): Default to yes for
14065 sparc*-*-linux* with glibc.
14067 2017-09-28 Joseph Myers <joseph@codesourcery.com>
14069 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
14070 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
14071 when creating .init_array and .fini_array sections with priority
14074 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
14077 * config/aarch64/aarch64.c
14078 (aarch64_builtin_support_vector_misalignment): Always return false
14079 when misalignment is unknown.
14081 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
14083 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
14084 this function to return false if the definition used by the swap
14085 instruction is artificial, or if the memory address from which the
14086 constant value is loaded is not represented by a base address held
14087 in a register or if the base address register is a frame or stack
14088 pointer. Additionally, return false if the base address of the
14089 loaded constant is a SYMBOL_REF but is not considered to be a
14091 (replace_swapped_load_constant): New function.
14092 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
14093 loaded constant vector with a load of a swapped constant vector.
14095 2017-09-27 Carl Love <cel@us.ibm.com>
14097 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
14098 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
14099 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
14102 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
14104 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
14105 first, always call autopref_rank_data otherwise.
14107 2017-09-27 Richard Biener <rguenther@suse.de>
14109 * graphite-scop-detection.c (find_scop_parameters): Move
14110 loop bound handling ...
14111 (gather_bbs::before_dom_children): ... here, avoiding the need
14112 to build scop_info->loop_nest.
14113 (record_loop_in_sese): Remove.
14114 * sese.h (sese_info_t::loop_nest): Remove.
14115 * sese.c (new_sese_info): Do not allocate loop_nest.
14116 (free_sese_info): Do not free loop_nest.
14118 2017-09-27 Jakub Jelinek <jakub@redhat.com>
14121 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
14122 lhs from calls if the lhs has addressable type.
14124 2017-09-27 Richard Biener <rguenther@suse.de>
14126 * graphite.h (scop::max_alias_set): New member.
14127 * graphite-scop-detection.c: Remove references to non-existing
14128 --param in comments.
14129 (build_alias_sets): Record the maximum alias set used for drs.
14130 (build_scops): Support zero as unlimited for
14131 --param graphite-max-arrays-per-scop.
14132 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
14133 and inline into ...
14134 (build_poly_sr_1): ... here. Compute alias set based on the
14135 maximum alias set used for drs rather than
14136 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
14138 2017-09-27 Richard Biener <rguenther@suse.de>
14140 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
14141 --param loop-block-tile-size=0 to disable tiling.
14143 2017-09-27 Richard Biener <rguenther@suse.de>
14145 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
14146 (graphite-max-nb-scop-params): Document special value zero.
14147 * domwalk.h (dom_walker::STOP): New symbolical constant.
14148 (dom_walker::dom_walker): Add optional parameter for bb to
14150 (dom_walker::~dom_walker): Declare.
14151 (dom_walker::before_dom_children): Document STOP return value.
14152 (dom_walker::m_user_bb_to_rpo): New member.
14153 (dom_walker::m_bb_to_rpo): Likewise.
14154 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
14155 mapping here if not provided by the user.
14156 (dom_walker::~dom_walker): Free bb to RPO mapping if not
14157 provided by the user.
14158 (dom_walker::STOP): Define.
14159 (dom_walker::walk): Do not compute bb to RPO mapping here.
14160 Support STOP return value from before_dom_children to stop
14162 * graphite-optimize-isl.c (optimize_isl): If the schedule
14163 is the same still generate code if -fgraphite-identity
14164 or -floop-parallelize-all are given.
14165 * graphite-scop-detection.c: Include cfganal.h.
14166 (gather_bbs::gather_bbs): Get and pass through bb to RPO
14168 (gather_bbs::before_dom_children): Return STOP for BBs
14170 (build_scops): Compute bb to RPO mapping and pass it to
14171 the domwalk. Treat --param graphite-max-nb-scop-params=0
14172 as not limiting the number of params.
14173 * graphite.c (graphite_initialize): Remove limit on the
14174 number of basic-blocks in a function.
14175 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
14176 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
14177 default value of 10.
14179 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14181 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
14182 Adjust code to eliminate needing to do the shift right 32-bits
14183 operation after XSCVDPSPN.
14185 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14187 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
14188 ((X / Y) != 0 -> X >= Y): Likewise.
14190 2017-09-26 Carl Love <cel@us.ibm.com>
14192 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
14193 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
14194 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
14195 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
14196 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
14197 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
14198 definitions and overloading.
14199 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
14200 statement for P9V_BUILTIN_XST_LEN_R.
14201 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
14202 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
14203 define_expand and define_insn for the instructions and builtins.
14204 * doc/extend.texi: Update the built-in documentation file for the new
14205 built-in functions.
14206 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
14207 define_insn for the instructions
14209 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
14212 * gcc/config/netbsd-protos.h: New file.
14213 * gcc/config/netbsd.c: New file.
14214 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
14215 * gcc/config/t-netbsd: New file.
14216 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
14217 (tmake_file) Add t-netbsd.
14218 (extra_objs) Add netbsd.o.
14220 2017-09-26 Janus Weil <janus@gcc.gnu.org>
14224 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
14226 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
14228 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
14229 sign extension from a vector register to a GPR by doing a 32-bit
14230 direct move and then an EXTSW.
14231 (extendsi<mode>2 splitter): Likewise.
14232 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
14233 right or vector extract after doing XSCVDPSPN. Use
14234 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
14236 (movdi_from_sf_zero_ext): Likewise.
14237 (reload_gpr_from_vsxsf): Likewise.
14238 (p8_mfvsrd_4_disf): Delete, no longer used.
14239 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
14240 then needing to move the SFmode to a GPR to use the XSCVDPSP
14241 instruction instead of FRSP and XSCVDPSPN.
14242 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
14243 it is adjacent to the other XSCVSPDP insns.
14244 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
14245 SFmode to be in traditional Altivec registers.
14246 (vsx_xscvdpspn): Eliminate useless alternative constraint.
14247 (vsx_xscvspdpn): Likewise.
14248 (vsx_xscvspdpn_scalar): Likewise.
14250 2017-09-26 Martin Jambor <mjambor@suse.cz>
14252 * tree-sra.c (compare_access_positions): Put integral types first,
14253 stabilize sorting of integral types, remove conditions putting
14254 non-full-precision integers last.
14255 (sort_and_splice_var_accesses): Disable scalarization if a
14256 non-integert would be represented by a non-full-precision integer.
14258 2017-09-26 Joseph Myers <joseph@codesourcery.com>
14260 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
14261 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
14262 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
14263 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
14264 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
14265 conditionals inside the function instead of around it. Call
14266 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
14267 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
14269 2017-09-26 Richard Biener <rguenther@suse.de>
14271 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
14273 (scop_detection::build_scop_breadth): ... this. Removed.
14274 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
14275 (scop_detection::harmful_stmt_in_bb): Likewise.
14276 (scop_detection::graphite_can_represent_stmt): Likewise.
14277 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
14278 (scop_detection::can_represent_loop): Remove recursion, fold in ...
14279 (scop_detection::can_represent_loop_1): ... this. Removed.
14280 (scop_detection::harmful_loop_in_region): Simplify after inlining
14281 the above and remove more quadraticness.
14282 (build_scops): Adjust.
14283 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
14286 2017-09-26 Jakub Jelinek <jakub@redhat.com>
14289 * config/i386/i386.c (ix86_print_operand_address_as): Only test
14290 REGNO (base) == SP_REG if base is a REG.
14292 PR middle-end/35691
14293 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
14294 if it is different SSA_NAME.
14295 (optimize_range_tests_cmp_bitwise): New function.
14296 (optimize_range_tests): Call it.
14298 2017-09-26 Richard Biener <rguenther@suse.de>
14300 PR tree-optimization/82321
14301 * graphite.c (canonicalize_loop_closed_ssa): Properly check
14302 for the def being inside the loop.
14304 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14306 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
14308 * config/s390/s390-builtins.def: Fix constraint on op4.
14310 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14312 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
14313 independent expanders.
14314 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
14315 ("vec_ordered", "vec_unordered"): New expanders.
14317 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14319 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
14322 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14324 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
14325 vec_unpacks_lo_v16qi.
14326 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
14328 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14330 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
14331 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
14332 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
14334 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14336 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
14338 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
14340 ("*vec_slb<mode>"): New insn pattern.
14341 ("vec_shr_<mode>"): New expander.
14342 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
14343 and force the shift count operand to V16QImode.
14344 ("vec_srb<mode>"): Set shift count mode to V16QI.
14346 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14348 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
14349 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
14350 ("vec_widen_smult_hi_<mode>"): New expander definitions.
14352 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
14355 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
14357 2017-09-26 Richard Biener <rguenther@suse.de>
14359 PR tree-optimization/82320
14360 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
14363 2017-09-25 Jeff Law <law@redhat.com>
14365 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
14366 prototype for new argument.
14367 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
14368 mostly extracted from rs6000_emit_allocate_stack.
14369 (rs6000_emit_probe_stack_range_stack_clash): New function.
14370 (rs6000_emit_allocate_stack): Call
14371 rs6000_emit_probe_stack_range_stack_clash as needed.
14372 (rs6000_emit_probe_stack_range): Add additional argument
14373 to call to gen_probe_stack_range{si,di}.
14374 (output_probe_stack_range): New.
14375 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
14376 (output_probe_stack_range_stack_clash): New.
14377 (rs6000_emit_prologue): Emit notes into dump file as requested.
14378 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
14379 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
14380 Add additional operand and pass it to output_probe_stack_range.
14382 2017-09-25 Bin Cheng <bin.cheng@arm.com>
14384 PR tree-optimization/82163
14385 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
14386 (checking_verify_loop_closed_ssa): New parameter.
14387 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
14388 (check_loop_closed_ssa_stmt): Delete.
14389 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
14390 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
14391 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
14394 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
14396 * brig-builtins.def: Treat HSAIL barrier builtins as
14397 setjmp/longjump style functions.
14399 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14401 * target.def (constant_alignment): New hook.
14402 * defaults.h (CONSTANT_ALIGNMENT): Delete.
14403 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
14404 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
14405 * doc/tm.texi: Regenerate.
14406 * targhooks.h (default_constant_alignment): Declare.
14407 (constant_alignment_word_strings): Likewise.
14408 * targhooks.c (default_constant_alignment): New function.
14409 (constant_alignment_word_strings): Likewise.
14410 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
14411 instead of CONSTANT_ALIGNMENT.
14412 * varasm.c (align_variable, get_variable_align, build_constant_desc)
14413 (force_const_mem): Likewise.
14414 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
14415 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
14416 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
14417 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14418 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
14420 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
14421 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14422 constant_alignment_word_strings.
14423 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
14424 (CONSTANT_ALIGNMENT): Likewise.
14425 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14426 (arm_constant_alignment): New function.
14427 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
14428 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14429 constant_alignment_word_strings.
14430 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
14431 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14432 constant_alignment_word_strings.
14433 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
14434 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14435 (cris_constant_alignment): New function.
14436 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
14437 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14438 (epiphany_constant_alignment): New function.
14439 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
14440 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14441 constant_alignment_word_strings.
14442 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
14443 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14444 constant_alignment_word_strings.
14445 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
14446 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14447 constant_alignment_word_strings.
14448 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
14449 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
14450 * config/i386/i386.c (ix86_constant_alignment): Make static.
14451 Use the same interface as the target hook.
14452 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14453 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
14454 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14455 constant_alignment_word_strings.
14456 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
14457 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
14458 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14459 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
14460 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14461 constant_alignment_word_strings.
14462 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
14463 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14464 constant_alignment_word_strings.
14465 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
14466 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14467 constant_alignment_word_strings.
14468 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
14469 * config/microblaze/microblaze.c (microblaze_constant_alignment):
14471 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14472 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
14473 * config/mips/mips.c (mips_constant_alignment): New function.
14474 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14475 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
14476 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
14477 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14478 (mmix_constant_alignment): Make static. Use the same interface
14479 as the target hook.
14480 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
14481 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14482 constant_alignment_word_strings.
14483 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
14484 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14485 constant_alignment_word_strings.
14486 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
14487 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14488 constant_alignment_word_strings.
14489 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
14490 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14491 (rs6000_constant_alignment): New function.
14492 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
14493 * config/riscv/riscv.c (riscv_constant_alignment): New function.
14494 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14495 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
14496 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14497 (rs6000_constant_alignment): New function.
14498 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
14499 * config/s390/s390.c (s390_constant_alignment): New function.
14500 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14501 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
14502 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14503 constant_alignment_word_strings.
14504 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
14505 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14506 (sparc_constant_alignment): New function.
14507 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
14508 * config/spu/spu.c (spu_constant_alignment): New function.
14509 (TARGET_CONSTANT_ALIGNMENT): Redefine.
14510 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
14511 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14512 constant_alignment_word_strings.
14513 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
14514 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14515 constant_alignment_word_strings.
14516 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
14517 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
14518 constant_alignment_word_strings.
14519 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
14520 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14521 (visium_constant_alignment): New function.
14522 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
14523 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
14524 (xtensa_constant_alignment): New function.
14525 * system.h (CONSTANT_ALIGNMENT): Poison.
14527 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
14529 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
14530 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
14531 (rs6000_builtin_valid_without_lhs): New helper function.
14532 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14533 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
14535 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
14537 * target.h (vec_perm_indices): Use unsigned short rather than
14539 (auto_vec_perm_indices): Likewise.
14540 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
14541 Use unsigned int rather than unsigned char.
14542 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
14544 2017-09-25 Richard Biener <rguenther@suse.de>
14546 * cfgloop.h (sort_sibling_loops): Declare.
14547 * cfgloop.c (sort_sibling_loops_cmp): New helper.
14548 (sort_sibling_loops): New function sorting the sibling loop list
14550 * graphite.c (graphite_transform_loops): Sort sibling loops.
14552 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
14554 * target.def (vec_perm_const_ok): Change sel parameter to
14556 * optabs-query.c (can_vec_perm_p): Update accordingly.
14557 * doc/tm.texi: Regenerate.
14558 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
14559 auto_vec_perm_indices and remove separate nelt field.
14560 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
14561 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
14562 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
14563 (aarch64_expand_vec_perm_const): Update accordingly.
14564 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
14565 to vec_perm_indices.
14566 * config/arm/arm.c (expand_vec_perm_d): Change perm to
14567 auto_vec_perm_indices and remove separate nelt field.
14568 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
14569 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
14570 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
14572 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
14573 to vec_perm_indices.
14574 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
14575 sel to vec_perm_indices.
14576 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
14577 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
14578 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
14580 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
14583 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
14586 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
14587 on the FUNCTION_DECL function context if it has a DIE that is a
14590 2017-09-25 Richard Biener <rguenther@suse.de>
14592 PR tree-optimization/82285
14593 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
14596 2017-09-25 Tom de Vries <tom@codesourcery.com>
14600 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
14603 2017-09-25 Richard Biener <rguenther@suse.de>
14605 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
14606 ISL errors other than isl_error_quota happen. Dump if the
14607 schedule is the same.
14608 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
14609 errors instead of aborting inside ISL.
14611 2017-09-25 Iain Sandoe <iain@codesourcery.com>
14614 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
14615 of libgcc_eh for m64.
14616 * config/i386/darwin64.h: Likewise.
14618 2017-09-25 Richard Biener <rguenther@suse.de>
14620 PR middle-end/82144
14621 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
14622 attribute for incomplete types nor twice for complete ones.
14624 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
14627 * config/i386/i386.c (ix86_print_operand_address_as): Encode
14628 %esp as %rsp to avoid 0x67 prefix if there is no index or base
14631 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
14634 * config/i386/i386.opt (mprefer-avx256): Use
14635 ix86_target_flags variable.
14636 * config/i386/i386.c (ix86_target_string): Move
14637 -mprefer-avx256 to flag2_opts.
14639 2017-09-22 Jakub Jelinek <jakub@redhat.com>
14641 PR middle-end/35691
14642 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
14643 and x != -1 | y != -1 into (x & y) != -1.
14645 2017-09-22 Steve Ellcey <sellcey@cavium.com>
14647 * config.gcc: Add new case statement to set
14648 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
14649 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
14650 s390x-*-linux* case statements. Added aarch64 to the list of
14651 supported architectures.
14653 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14655 PR tree-optimization/82289
14656 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
14657 STMT_VINFO_RELEVANT_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 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
14664 for VR_RANGE only; don't allow VR_ANTI_RANGE.
14665 (extract_range_from_binary_expr_1): Don't call
14666 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
14668 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14669 Alan Hayward <alan.hayward@arm.com>
14670 David Sherwood <david.sherwood@arm.com>
14672 * target.def (preferred_vector_alignment): New hook.
14673 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
14675 * doc/tm.texi: Regenerate.
14676 * targhooks.h (default_preferred_vector_alignment): Declare.
14677 * targhooks.c (default_preferred_vector_alignment): New function.
14678 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
14680 (DR_TARGET_ALIGNMENT): New macro.
14681 (aligned_access_p): Update commentary.
14682 (vect_known_alignment_in_bytes): New function.
14683 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
14685 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
14686 Calculate the misalignment based on the target alignment rather than
14688 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
14689 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
14690 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
14691 the target alignment, rather than masking the element misalignment
14692 with the number of elements in a vector. Also use the target
14693 alignment when calculating the maximum number of peels.
14694 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
14695 instead of TYPE_ALIGN_UNIT.
14696 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
14697 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
14698 (vect_create_addr_base_for_vector_ref): Update call accordingly.
14699 (vect_create_data_ref_ptr): Likewise.
14700 (vect_setup_realignment): Realign by ANDing with
14701 -DR_TARGET_MISALIGNMENT.
14702 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
14703 the number of peels based on DR_TARGET_ALIGNMENT.
14704 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
14705 with the guaranteed alignment boundary when deciding whether
14707 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
14708 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
14709 (ensure_base_align): Remove stmt_info parameter. Get the
14710 target base alignment from DR_TARGET_ALIGNMENT.
14711 (vectorizable_store): Update call accordingly. Interpret
14712 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
14714 (vectorizable_load): Likewise.
14716 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
14717 Alan Hayward <alan.hayward@arm.com>
14718 David Sherwood <david.sherwood@arm.com>
14720 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
14721 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
14722 (vect_enhance_data_refs_alignment): Likewise.
14724 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14726 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
14727 error. Only quit immediately if parsing is complete.
14728 (BEGIN): Initialize fatal_err and parse_done.
14729 (begin fpu, end fpu): Check number of arguments.
14730 (begin arch, end arch): Likewise.
14731 (begin cpu, end cpu): Likewise.
14732 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
14733 (optalias): Likewise.
14735 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
14737 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
14738 * config/arm/arm-isa.h: Delete. Move definitions to ...
14739 * arm-cpus.in: ... here. Use new feature and fgroup values.
14740 * config/arm/arm.c (arm_option_override): Use lower case for feature
14742 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
14743 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
14744 * config/arm/parsecpu.awk (END): Add new command 'isa'.
14746 (print_isa_bits_for): New function.
14747 (gen_isa): New function.
14748 (gen_comm_data): Use print_isa_bits_for.
14749 (define feature): New keyword.
14750 (define fgroup): New keyword.
14751 * config/arm/t-arm (TM_H): Remove.
14752 (GTM_H): Add arm-isa.h.
14753 (arm-isa.h): Add rule to generate file.
14754 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
14755 case for feature bit names.
14757 2017-09-22 Richard Biener <rguenther@suse.de>
14759 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
14761 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
14762 print of no dependency loops ...
14763 * graphite.c (graphite_transform_loops): ... here.
14764 (canonicalize_loop_closed_ssa_form): Work from inner to outer
14766 (same_close_phi_node, remove_duplicate_close_phi,
14767 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
14768 (canonicalize_loop_closed_ssa): ... here and simplify.
14769 * graphite-optimize-isl.c: Include tree-vectorizer.h.
14770 (optimize_isl): Use dump_printf_loc to tell when we stopped
14771 optimizing because of an ISL timeout.
14773 2017-09-22 Richard Biener <rguenther@suse.de>
14775 PR tree-optimization/82291
14776 * tree-if-conv.c (predicate_mem_writes): Make sure to
14777 remove writes in blocks predicated with false.
14779 2017-09-22 Richard Biener <rguenther@suse.de>
14781 * sese.c: Include cfganal.h.
14782 (if_region_set_false_region): Remove.
14783 (create_if_region_on_edge): Likewise.
14784 (move_sese_in_condition): Re-implement without destroying
14787 2017-09-22 Richard Biener <rguenther@suse.de>
14789 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
14790 Verify both BBs contain loop PHI nodes before dispatching to
14791 copy_loop_phi_args.
14792 (graphite_regenerate_ast_isl): Do not recompute dominators,
14793 do not verify three times. Restructure for clarity.
14794 * graphite-scop-detection.c (same_close_phi_node,
14795 remove_duplicate_close_phi, make_close_phi_nodes_unique,
14796 defined_in_loop_p, canonicalize_loop_closed_ssa,
14797 canonicalize_loop_closed_ssa_form): Simplify, remove excess
14798 checking and SSA rewrite, move to ...
14799 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
14800 (graphite_initialize): Do not pass in ctx, do not reset the
14801 SCEV cache, compute only dominators.
14802 (graphite_transform_loops): Allocate ISL ctx after
14803 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
14804 Maintain post-dominators only around build_scops.
14805 * sese.c (if_region_set_false_region): Make static. Free
14806 and recompute dominators.
14807 (move_sese_in_condition): Assert we don't get called with
14808 post-dominators computed.
14809 * sese.h (if_region_set_false_region): Remove.
14811 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
14813 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
14814 mode attribute for TARGET_AVX512VL.
14816 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
14818 * config/i386/i386.opt (mprefer-avx256): New option.
14819 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
14821 (ix86_preferred_simd_mode): Return 256-bit AVX modes
14822 for TARGET_PREFER_AVX256.
14823 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
14825 2017-09-21 Jeff Law <law@redhat.com>
14827 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
14828 Fix dump output if the only stack space is for pushed registers.
14830 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14832 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
14835 2017-09-21 Martin Sebor <msebor@redhat.com>
14838 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
14839 code (in C++) or code that triggers warnings.
14841 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
14843 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
14845 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
14847 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
14848 * sched-rgn.c: Ditto.
14849 * sel-sched-ir.c: Ditto.
14851 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
14853 * toplev.h (set_random_seed): Adjust return type.
14854 * toplev.c (init_local_tick): Move eager initialization of random_seed
14855 to get_random_seed. Adjust comment.
14856 (init_random_seed): Inline to get_random_seed, delete.
14857 (get_random_seed): Initialize random_seed lazily.
14858 (set_random_seed): Do not return previous value.
14859 (print_switch_value): Do not call get_random_seed.
14861 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
14863 * cgraph.c (delete_function_version): New, broken out from...
14864 (cgraph_node::delete_function_version): ...here. Rename to
14865 cgraph_node::delete_function_version_by_decl. Update all uses.
14866 (cgraph_node::remove): Call delete_function_version.
14868 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14871 * tree-inline.c (expand_call_inline): Emit clobber stmts for
14872 VAR_DECLs to which addressable non-volatile parameters are mapped
14873 and for id->retvar after the return value assignment. Clear
14874 id->retval and id->retbnd after inlining.
14876 2017-09-21 Richard Biener <rguenther@suse.de>
14878 PR tree-optimization/82276
14879 PR tree-optimization/82244
14880 * tree-vrp.c (build_assert_expr_for): Set
14881 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
14883 (remove_range_assertions): Revert earlier change.
14885 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
14888 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
14890 2017-09-21 Richard Biener <rguenther@suse.de>
14892 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
14893 Restore valid IL after code generation errors.
14894 * graphite.c (graphite_transform_loops): Diagnose code
14895 generation issues as MSG_MISSED_OPTIMIZATION and continue
14896 with processing SCOPs.
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 * calls.c (compute_argument_addresses): Use simplify_gen_binary
14903 rather than choosing between plus_constant and gen_rtx_<CODE>.
14904 * expr.c (emit_push_insn): Likewise.
14905 (expand_expr_real_2): Likewise.
14907 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14908 Alan Hayward <alan.hayward@arm.com>
14909 David Sherwood <david.sherwood@arm.com>
14911 * loop-unroll.c (split_iv): Call copy_rtx on the step.
14913 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14914 Alan Hayward <alan.hayward@arm.com>
14915 David Sherwood <david.sherwood@arm.com>
14917 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
14918 calling tree_to_uhwi.
14920 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14921 Alan Hayward <alan.hayward@arm.com>
14922 David Sherwood <david.sherwood@arm.com>
14924 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
14925 INTEGER_CST rather than a negative test for ADDR_EXPR.
14927 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
14928 Alan Hayward <alan.hayward@arm.com>
14929 David Sherwood <david.sherwood@arm.com>
14931 * tree-vrp.c (extract_range_from_binary_expr_1): Check
14932 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
14934 2017-09-21 Richard Biener <rguenther@suse.de>
14936 PR tree-optimization/71351
14937 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
14938 graphite_create_new_loop_guard): Remove, fold remaining parts
14940 (translate_isl_ast_node_for): ... here and simplify.
14942 2017-09-21 Jakub Jelinek <jakub@redhat.com>
14945 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
14946 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
14947 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
14948 alternative always use QI mode, for -Os imov (=R,R) alternative
14949 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
14952 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14953 Jeff Law <law@redhat.com>
14955 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
14956 (allocate_stack_space): New function, partially extracted from
14957 s390_emit_prologue.
14958 (s390_emit_prologue): Track offset to most recent stack probe.
14959 Code to allocate space moved into allocate_stack_space.
14960 Dump actions when no stack is allocated.
14961 (s390_prologue_plus_offset): New function.
14962 (s390_emit_stack_probe): Likewise.
14964 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
14966 * common.opt (Wa, Wl, Wp, g, gz=): Add
14968 (gno-column-info): Remove.
14969 (gcolumn-info): Drop RejectNegative.
14970 (gno-): New prefix.
14971 (gno-record-gcc-switches): Remove.
14972 (grecord-gcc-switches): Drop RejectNegative.
14973 (gno-split-dwarf): Remove.
14974 (gsplit-dwarf): Drop RejectNegative.
14975 (gno-strict-dwarf): Remove.
14976 (gstrict-dwarf): Drop RejectNegative.
14977 * config/darwin.opt (gfull, gused): Add RejectNegative.
14978 * dwarf2out.c (gen_producer_string): Drop
14979 gno-record-gcc-switches handler.
14980 * optc-gen.awk: Add g to prefixes with negative forms.
14981 * opts-common.c (remapping_prefix_p): New.
14982 (find_opt): Check it.
14983 (generate_canonical_option): Test g prefix.
14984 (option_map): Add -gno- mapping.
14985 (add_misspelling_candidates): Check remapping_prefix_p.
14987 2017-09-20 Jeff Law <law@redhat.com>
14989 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
14990 thinko in stack clash protection support.
14992 * explow.c (compute_stack_clash_protection_loop_data): Use
14993 CONST_INT_P instead of explicit test. Verify object is a
14994 CONST_INT_P before looking at INTVAL.
14995 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
14996 instead of explicit test.
14998 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
15001 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
15002 address instead of to r1 and r11.
15004 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
15006 * config.gcc: Support "knm".
15007 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
15008 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
15010 * config/i386/i386.c (m_KNM): Define.
15011 (processor_target_table): Add "knm".
15013 (ix86_option_override_internal): Add "knm".
15014 (ix86_issue_rate): Add PROCESSOR_KNM.
15015 (ix86_adjust_cost): Ditto.
15016 (ia32_multipass_dfa_lookahead): Ditto.
15017 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
15018 (fold_builtin_cpu): Add M_INTEL_KNM.
15019 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
15020 (processor_type): Add PROCESSOR_KNM.
15021 * config/i386/x86-tune.def: Add m_KNM.
15022 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
15024 2017-09-20 Richard Biener <rguenther@suse.de>
15026 PR tree-optimization/80213
15027 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
15028 are allowed in empty BBs as well.
15029 (canonicalize_loop_closed_ssa): Also look for other complex
15031 (scop_detection::get_sese): Include the loop-closed PHI block
15033 (scop_detection::merge_sese): Remove code adding extra blocks.
15034 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
15035 (build_scops): Assert the final returned scop is invalid.
15037 2017-09-20 Richard Biener <rguenther@suse.de>
15039 PR tree-optimization/82264
15040 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
15042 (vn_phi_lookup): Likewise.
15043 (vn_phi_insert): Likewise.
15045 2017-09-20 Jakub Jelinek <jakub@redhat.com>
15047 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
15048 that fits into uhwi or shwi, add DW_AT_const_value regardless
15049 of early_dwarf without going through RTL, using add_AT_unsigned
15052 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
15053 (DEBUG_LTO_ABBREV_SECTION): Likewise.
15054 (DEBUG_LTO_MACINFO_SECTION): Likewise.
15055 (DEBUG_MACRO_SECTION): Likewise.
15056 (DEBUG_LTO_MACRO_SECTION): Likewise.
15057 (DEBUG_STR_DWO_SECTION): Likewise.
15058 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
15059 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
15060 (DEBUG_LTO_DWO_LINE_SECTION): Define.
15061 (DEBUG_LTO_LINE_STR_SECTION): Define.
15062 (init_sections_and_labels): Initialize debug_line_str_section
15063 variable. Initialize debug_loc_section for -gdwarf-5 to
15064 DEBUG_LOCLISTS_SECTION. Formatting fixes.
15066 2017-09-20 Richard Biener <rguenther@suse.de>
15068 * graphite-sese-to-poly.c (extract_affine): Properly handle
15069 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
15071 2017-09-20 Richard Biener <rguenther@suse.de>
15073 PR tree-optimization/81373
15074 * graphite-scop-detection.c (build_cross_bb_scalars_def):
15075 Force SESE live-out defs to be handled even if they are
15078 2017-09-19 Jeff Law <law@redhat.com>
15080 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
15081 nothing for stack adjustments with REG_STACK_CHECK.
15082 * sched-deps.c (parse_add_or_inc): Reject insns with
15083 REG_STACK_CHECK from dependency breaking.
15084 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
15085 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
15086 * reg-notes.def (STACK_CHECK): New note.
15088 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
15089 (ix86_expand_prologue): Dump stack clash info as needed.
15090 Call ix86_adjust_stack_and_probe_stack_clash as needed.
15092 * function.c (dump_stack_clash_frame_info): New function.
15093 * function.h (dump_stack_clash_frame_info): Prototype.
15094 (enum stack_clash_probes): New enum.
15096 * config/alpha/alpha.c (alpha_expand_prologue): Also check
15097 flag_stack_clash_protection.
15098 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
15099 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
15100 (arm_frame_pointer_required): Likewise.
15101 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
15102 (ia64_expand_prologue): Likewise.
15103 * config/mips/mips.c (mips_expand_prologue): Likewise.
15104 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
15105 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15106 (sparc_flat_expand_prologue): Likewise.
15107 * config/spu/spu.c (spu_expand_prologue): Likewise.
15109 * explow.c: Include "params.h".
15110 (anti_adjust_stack_and_probe_stack_clash): New function.
15111 (get_stack_check_protect): Likewise.
15112 (compute_stack_clash_protection_loop_data): Likewise.
15113 (emit_stack_clash_protection_loop_start): Likewise.
15114 (emit_stack_clash_protection_loop_end): Likewise.
15115 (allocate_dynamic_stack_space): Use get_stack_check_protect.
15116 Use anti_adjust_stack_and_probe_stack_clash.
15117 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
15118 (emit_stack_clash_protection_loop_start): Likewise.
15119 (emit_stack_clash_protection_loop_end): Likewise.
15120 * rtl.h (get_stack_check_protect): Prototype.
15121 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
15122 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
15123 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
15125 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
15127 * doc/tm.texi: Rebuilt.
15128 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
15129 get_stack_check_protect.
15130 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
15131 * config/arm/arm.c (arm_expand_prologue): Likewise.
15132 (arm_frame_pointer_required): Likewise.
15133 * config/i386/i386.c (ix86_expand_prologue): Likewise.
15134 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
15135 * config/mips/mips.c (mips_expand_prologue): Likewise.
15136 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
15137 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
15138 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
15139 (sparc_flat_expand_prologue): Likewise.
15141 * common.opt (-fstack-clash-protection): New option.
15142 * flag-types.h (enum stack_check_type): Note difference between
15143 -fstack-check= and -fstack-clash-protection.
15144 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
15145 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
15146 * toplev.c (process_options): Issue warnings/errors for cases
15147 not handled with -fstack-clash-protection.
15148 * doc/invoke.texi (-fstack-clash-protection): Document new option.
15149 (-fstack-check): Note additional problem with -fstack-check=generic.
15150 Note that -fstack-check is primarily for Ada and refer users
15151 to -fstack-clash-protection for stack-clash-protection.
15152 Document new params for stack clash protection.
15154 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
15156 * config/i386/i386.c (ix86_split_long_move): Do not handle
15157 address used for LEA in a special way.
15159 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
15161 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
15163 2017-09-19 Martin Sebor <msebor@redhat.com>
15166 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
15167 of incompatible types.
15169 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
15171 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
15172 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
15173 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15174 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
15176 2017-09-19 Richard Biener <rguenther@suse.de>
15178 PR tree-optimization/82244
15179 * tree-vrp.c (remove_range_assertions): Do not propagate
15180 a constant to abnormals but replace the assert with a copy.
15182 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
15184 PR rtl-optimization/57878
15185 PR rtl-optimization/68988
15186 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
15187 avoidance test involving non_reload_pseudos. Move frequency test
15188 below the general fragmentation avoidance test.
15190 2017-09-19 Richard Biener <rguenther@suse.de>
15192 PR tree-optimization/69728
15193 * graphite-sese-to-poly.c (schedule_error): New global.
15194 (add_loop_schedule): Handle empty domain by failing the
15196 (build_original_schedule): Handle schedule_error.
15198 2017-09-19 Richard Biener <rguenther@suse.de>
15200 * graphite-scop-detection.c (scop_detection::can_represent_loop):
15201 Do not iterate to sibling loops but only to siblings of inner
15204 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
15207 * config/m68k/m68k.md (moveq feeding equality comparison): Check
15208 that the registers are different.
15210 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
15212 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
15213 to processor_model and "amdfam17h" to arch_names_table.
15214 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
15216 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15219 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
15221 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15222 Alan Hayward <alan.hayward@arm.com>
15223 David Sherwood <david.sherwood@arm.com>
15225 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
15227 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
15229 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
15230 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
15231 vect_schedule_slp_instance.
15232 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
15233 Update call to vect_slp_analyze_node_operations. Simplify return
15235 (vect_slp_analyze_bb_1): Update call accordingly.
15236 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
15237 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
15238 (vect_schedule_slp): Update call accordingly.
15240 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15241 Alan Hayward <alan.hayward@arm.com>
15242 David Sherwood <david.sherwood@arm.com>
15244 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
15245 with types that aren't in fact scalar.
15247 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15249 * tree-vect-slp.c (vect_record_max_nunits): New function,
15251 (vect_build_slp_tree_1): ...here.
15252 (vect_build_slp_tree_2): Call it for phis too.
15254 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15256 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
15257 to vect_get_vec_def_for_operand when getting the mask operand.
15259 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15260 Alan Hayward <alan.hayward@arm.com>
15261 David Sherwood <david.sherwood@arm.com>
15263 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
15266 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15267 Alan Hayward <alan.hayward@arm.com>
15268 David Sherwood <david.sherwood@arm.com>
15270 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
15271 calculation for vector booleans.
15273 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
15274 Alan Hayward <alan.hayward@arm.com>
15275 David Sherwood <david.sherwood@arm.com>
15277 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
15279 (vect_transform_stmt): ...here.
15280 (vect_analyze_stmt): Use it instead of calling
15281 vectorizable_live_operation directly.
15283 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
15285 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
15286 non-SIMT targets in acc vector loops.
15288 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
15290 * configure.ac: Add arc and check if assembler supports gdwarf2.
15291 * configure: Regenerate.
15293 2017-09-18 Richard Biener <rguenther@suse.de>
15295 PR tree-optimization/82220
15296 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
15297 epilogue niters from the min_profitable_iters compute.
15299 2017-09-18 Jakub Jelinek <jakub@redhat.com>
15302 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
15303 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
15304 (ix86_init_pic_reg): Revert 2017-09-01 changes.
15306 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
15309 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
15310 switching to a new text section.
15312 2017-09-18 Richard Biener <rguenther@suse.de>
15314 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
15316 (build_alias_set): Reject aliases with no access function.
15318 2017-09-18 Richard Biener <rguenther@suse.de>
15320 PR tree-optimization/79622
15321 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
15323 (build_cross_bb_scalars_use): Likewise.
15325 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
15327 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
15329 2017-09-18 Alan Modra <amodra@gmail.com>
15332 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
15333 stack_pointer_rtx for count 0. Update comments. Break up
15334 large rtl expression.
15336 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
15338 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
15339 Increase to 20 bytes.
15340 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
15341 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
15342 or avx version of the stub.
15344 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
15347 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
15348 compute the minimum stack alignment. Also update preferred stack
15349 boundary for leaf functions.
15351 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
15353 PR tree-optimization/82228
15354 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
15357 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15359 * common/config/nds32/nds32-common.c
15360 (nds32_option_optimization_table): Refine formatting.
15361 (nds32_option_optimization_table): Use -fsched-pressure and
15362 -fomit-frame-pointer for specific optimization level.
15364 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
15366 * config/nds32/nds32.c: Refine formatting and comments.
15367 * config/nds32/nds32.h: Likewise.
15368 * config/nds32/nds32.md: Likewise.
15369 * config/nds32/nds32-cost.c: Likewise.
15370 * config/nds32/nds32-isr.c: Likewise.
15371 * config/nds32/nds32-md-auxiliary.c: Likewise.
15372 * config/nds32/nds32-multiple.md: Likewise.
15373 * config/nds32/nds32-predicates.c: Likewise.
15375 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
15376 Jakub Jelinek <jakub@redhat.com>
15378 Add support for -std=c++2a.
15379 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
15381 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
15383 2017-09-15 Steve Ellcey <sellcey@cavium.com>
15386 * doc/extend.texi (Common Function Attributes): Add
15387 references to ARM, AArch64, and S/390 specific attributes.
15388 (Function Specific Option Pragmas): Add AArch64 and S/390
15389 to list of back ends that support the target pragma.
15391 2017-09-15 Nathan Sidwell <nathan@acm.org>
15393 * doc/standards.texi: Fix C++17 description. Update URLs for
15396 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
15398 * common.opt (Wcast-align=strict): New warning option.
15399 * doc/invoke.texi: Document -Wcast-align=strict.
15401 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
15403 * cgraph.h (cgraph_thunk_info): Add comments.
15404 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
15405 assert for VIRTUAL_* arguments stricter.
15407 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
15409 PR tree-optimization/71026
15410 * match.pd: Move RDIV patterns from fold-const.c
15411 * fold-const.c (distribute_real_division): Removed.
15412 (fold_binary_loc): Remove calls to distribute_real_divison.
15414 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15416 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
15417 c++1z and gnu++1z as deprecated. Change other references to
15418 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
15419 Change -Wc++1z-compat to -Wc++17-compat.
15420 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
15421 * dwarf2out.c (highest_c_language): Handle C++17.
15422 (gen_compile_unit_die): Likewise.
15424 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15426 PR rtl-optimization/82192
15427 * combine.c (make_extraction): Don't look through non-paradoxical
15428 SUBREGs or TRUNCATE if pos + len is or might be bigger than
15431 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15432 Alan Hayward <alan.hayward@arm.com>
15433 David Sherwood <david.sherwood@arm.com>
15435 * target.def (function_arg_offset): New hook.
15436 * targhooks.h (default_function_arg_offset): Declare.
15437 * targhooks.c (default_function_arg_offset): New function.
15438 * function.c (locate_and_pad_parm): Use
15439 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
15440 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
15441 (TARGET_FUNCTION_ARG_OFFSET): ...this.
15442 * doc/tm.texi: Regenerate.
15443 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
15444 * config/spu/spu.c (spu_function_arg_offset): New function.
15445 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
15446 * system.h (FUNCTION_ARG_OFFSET): Poison.
15448 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15449 Alan Hayard <alan.hayward@arm.com>
15450 David Sherwood <david.sherwood@arm.com>
15452 * target.def (truly_noop_truncation): New hook.
15453 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
15454 than TRULY_NOOP_TRUNCATION.
15455 * hooks.h (hook_bool_uint_uint_true): Declare.
15456 * hooks.c (hook_bool_uint_uint_true): New function.
15457 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
15458 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
15459 * doc/tm.texi: Regenerate.
15460 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
15461 rather than TRULY_NOOP_TRUNCATION in comments.
15462 (simplify_comparison): Likewise.
15463 (record_truncated_value): Likewise.
15464 * expmed.c (extract_bit_field_1): Likewise.
15465 (extract_split_bit_field): Likewise.
15466 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
15467 instead of TRULY_NOOP_TRUNCATION.
15468 * function.c (assign_parm_setup_block): Likewise.
15469 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
15470 * rtlhooks.c: Include target.h.
15471 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
15472 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
15473 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
15474 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
15475 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
15476 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
15477 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
15478 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
15479 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
15480 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
15481 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
15482 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
15483 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
15484 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
15485 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
15486 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
15487 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
15488 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
15489 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
15490 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
15491 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
15492 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
15493 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
15494 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
15495 * config/mips/mips.c (mips_truly_noop_truncation): New function.
15496 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15497 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15498 rather than TRULY_NOOP_TRUNCATION in comments.
15499 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
15500 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
15501 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
15502 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
15503 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
15504 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
15505 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
15506 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
15507 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
15508 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
15509 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
15510 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15511 rather than TRULY_NOOP_TRUNCATION in comments.
15512 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
15513 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
15514 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
15515 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
15516 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
15517 TRULY_NOOP_TRUNCATION condition.
15518 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
15519 (TRULY_NOOP_TRUNCATION): Delete.
15520 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
15521 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
15522 * config/spu/spu.c (spu_truly_noop_truncation): New function.
15523 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15524 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
15525 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
15526 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
15527 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
15528 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
15529 rather than TRULY_NOOP_TRUNCATION in comments.
15530 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
15531 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
15532 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
15533 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
15534 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
15535 * system.h (TRULY_NOOP_TRUNCATION): Poison.
15537 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
15540 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
15541 (*cmp_ior): Likewise.
15542 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
15543 (*ior_scc_scc_cmp): Likewise.
15544 (*and_scc_scc): Likewise.
15545 (*and_scc_scc_cmp): Likewise.
15547 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
15548 Alan Hayard <alan.hayward@arm.com>
15549 David Sherwood <david.sherwood@arm.com>
15551 * target.def (can_change_mode_class): New hook.
15552 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
15553 (hard_regno_nregs): Likewise.
15554 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
15555 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
15556 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
15557 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
15558 (LOAD_EXTEND_OP): Update accordingly.
15559 * doc/tm.texi: Regenerate.
15560 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
15561 CANNOT_CHANGE_MODE_CLASS.
15562 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
15563 (REG_CAN_CHANGE_MODE_P): ...this new macro.
15564 * combine.c (simplify_set): Update accordingly.
15565 * emit-rtl.c (validate_subreg): Likewise.
15566 * recog.c (general_operand): Likewise.
15567 * regcprop.c (mode_change_ok): Likewise.
15568 * reload1.c (choose_reload_regs): Likewise.
15569 (inherit_piecemeal_p): Likewise.
15570 * rtlanal.c (simplify_subreg_regno): Likewise.
15571 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
15572 instead of CANNOT_CHANGE_MODE_CLASS.
15573 (reload_cse_simplify_operands): Likewise.
15574 * reload.c (push_reload): Use targetm.can_change_mode_class
15575 instead of CANNOT_CHANGE_MODE_CLASS.
15576 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
15577 REG_CANNOT_CHANGE_MODE_P.
15578 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15579 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
15580 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15581 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15582 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15583 (arm_can_change_mode_class): New function.
15584 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15585 than CANNOT_CHANGE_MODE_CLASS in comments.
15586 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15587 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
15588 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
15589 (ix86_can_change_mode_class): ...this new function, inverting the
15590 sense of the return value.
15591 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15592 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15593 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15594 (ia64_can_change_mode_class): New function.
15595 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15596 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
15597 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
15598 (m32c_can_change_mode_class): ...this new function, inverting the
15599 sense of the return value.
15600 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15601 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15602 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
15603 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
15604 (mips_can_change_mode_class): ...this new function, inverting the
15605 sense of the return value.
15606 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15607 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15608 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15609 (msp430_can_change_mode_class): New function.
15610 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15611 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
15612 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15613 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15614 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15615 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
15616 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15617 (pa_cannot_change_mode_class): Replace with...
15618 (pa_can_change_mode_class): ...this new function, inverting the
15619 sense of the return value.
15620 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
15621 than CANNOT_CHANGE_MODE_CLASS in comments.
15622 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15623 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
15624 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15625 (pdp11_cannot_change_mode_class): Replace with...
15626 (pdp11_can_change_mode_class): ...this new function, inverting the
15627 sense of the return value.
15628 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15629 * config/powerpcspe/powerpcspe-protos.h
15630 (rs6000_cannot_change_mode_class_ptr): Delete.
15631 * config/powerpcspe/powerpcspe.c
15632 (rs6000_cannot_change_mode_class_ptr): Delete.
15633 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15634 (rs6000_option_override_internal): Assign to
15635 targetm.can_change_mode_class instead of
15636 rs6000_cannot_change_mode_class_ptr.
15637 (rs6000_cannot_change_mode_class): Replace with...
15638 (rs6000_can_change_mode_class): ...this new function, inverting the
15639 sense of the return value.
15640 (rs6000_debug_cannot_change_mode_class): Replace with...
15641 (rs6000_debug_can_change_mode_class): ...this new function.
15642 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15643 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
15644 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15645 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15646 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
15648 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
15649 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15650 (rs6000_option_override_internal): Assign to
15651 targetm.can_change_mode_class instead of
15652 rs6000_cannot_change_mode_class_ptr.
15653 (rs6000_cannot_change_mode_class): Replace with...
15654 (rs6000_can_change_mode_class): ...this new function, inverting the
15655 sense of the return value.
15656 (rs6000_debug_cannot_change_mode_class): Replace with...
15657 (rs6000_debug_can_change_mode_class): ...this new function.
15658 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15659 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
15660 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
15661 (s390_can_change_mode_class): ...this new function, inverting the
15662 sense of the return value.
15663 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15664 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15665 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
15666 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15667 (sh_cannot_change_mode_class): Replace with...
15668 (sh_can_change_mode_class): ...this new function, inverting the
15669 sense of the return value.
15670 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15671 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15672 (sparc_can_change_mode_class): New function.
15673 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15674 * config/spu/spu.c (spu_can_change_mode_class): New function.
15675 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15676 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
15677 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
15678 (visium_can_change_mode_class): New function.
15679 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
15681 2017-09-15 Richard Biener <rguenther@suse.de>
15683 PR tree-optimization/82217
15684 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
15685 but not undefined case.
15687 2017-09-15 Jakub Jelinek <jakub@redhat.com>
15690 * postreload.c (reload_cse_simplify_operands): Skip
15691 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
15693 2017-09-15 Richard Biener <rguenther@suse.de>
15695 PR tree-optimization/68823
15696 * graphite-scop-detection.c (build_alias_set): If we have a
15697 possible dependence check whether we can handle them by just
15698 looking at the DRs DR_ACCESS_FNs.
15699 (build_scops): If build_alias_set fails, fail the SCOP.
15701 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15703 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
15704 to support float128 built-in functions that require the ISA 3.0
15706 (BU_FLOAT128_3_HW): Likewise.
15707 (SQRTF128): Add support for the IEEE 128-bit square root and fma
15708 built-in functions.
15709 (FMAF128): Likewise.
15711 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
15712 support for built-in functions that need the ISA 3.0 IEEE 128-bit
15713 floating point instructions.
15714 (rs6000_invalid_builtin): Likewise.
15715 (rs6000_builtin_mask_names): Likewise.
15716 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
15717 (RS6000_BTM_FLOAT128_HW): Likewise.
15718 (RS6000_BTM_COMMON): Likewise.
15719 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
15721 * doc/extend.texi (RS/6000 built-in functions): Document the
15722 IEEE 128-bit floating point square root and fused multiply-add
15723 built-in functions.
15725 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
15727 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
15728 reg (r2) isn't in the set of registers defined in the prologue.
15730 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15731 Alan Hayward <alan.hayward@arm.com>
15732 David Sherwood <david.sherwood@arm.com>
15734 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
15735 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
15736 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
15737 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
15738 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
15740 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
15741 max_vectorization_factor.
15742 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
15744 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15745 Alan Hayward <alan.hayward@arm.com>
15746 David Sherwood <david.sherwood@arm.com>
15748 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
15749 (vect_worthwhile_without_simd_p): Declare.
15750 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
15751 (vectorizable_reduction): Use it.
15752 * tree-vect-stmts.c (vectorizable_shift): Likewise.
15753 (vectorizable_operation): Likewise.
15755 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15756 Alan Hayward <alan.hayward@arm.com>
15757 David Sherwood <david.sherwood@arm.com>
15759 * tree-vectorizer.h (vect_get_num_copies): New function.
15760 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
15761 * tree-vect-loop.c (vectorizable_reduction): Likewise.
15762 (vectorizable_induction): Likewise.
15763 (vectorizable_live_operation): Likewise.
15764 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
15765 (vectorizable_bswap): Likewise.
15766 (vectorizable_call): Likewise.
15767 (vectorizable_conversion): Likewise.
15768 (vectorizable_assignment): Likewise.
15769 (vectorizable_shift): Likewise.
15770 (vectorizable_operation): Likewise.
15771 (vectorizable_store): Likewise.
15772 (vectorizable_load): Likewise.
15773 (vectorizable_condition): Likewise.
15774 (vectorizable_comparison): Likewise.
15775 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
15777 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15778 Alan Hayward <alan.hayward@arm.com>
15779 David Sherwood <david.sherwood@arm.com>
15781 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
15782 of vect_init_vector.
15784 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15785 Alan Hayward <alan.hayward@arm.com>
15786 David Sherwood <david.sherwood@arm.com>
15788 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
15789 an inline wrapper that provides a location.
15790 (gimple_build_vector): Likewise.
15791 * gimple-fold.c (gimple_build_vector_from_val): New function.
15792 (gimple_build_vector): Likewise.
15793 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
15794 functions to build the initial value. Always return a gimple value.
15795 (get_initial_defs_for_reduction): Likewise. Only compute
15797 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
15798 vect_init_vector on the results from get_initial_def(s)_for_reduction.
15799 (vectorizable_induction): Use gimple_build_vector rather than
15802 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15803 Alan Hayward <alan.hayward@arm.com>
15804 David Sherwood <david.sherwood@arm.com>
15806 * target.h (vec_perm_indices): New typedef.
15807 (auto_vec_perm_indices): Likewise.
15808 * optabs-query.h: Include target.h
15809 (can_vec_perm_p): Take a vec_perm_indices *.
15810 * optabs-query.c (can_vec_perm_p): Likewise.
15811 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
15812 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15813 * tree-vect-generic.c (lower_vec_perm): Likewise.
15814 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
15815 (vect_grouped_load_supported): Likewise.
15816 (vect_shift_permute_load_chain): Likewise.
15817 (vect_permute_store_chain): Use auto_vec_perm_indices.
15818 (vect_permute_load_chain): Likewise.
15819 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
15820 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
15821 Update uses of can_vec_perm_p.
15822 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
15823 mode with a number of elements. Take a vec_perm_indices *.
15824 (vect_create_epilog_for_reduction): Update accordingly.
15825 Use auto_vec_perm_indices.
15826 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
15827 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
15828 (vect_transform_slp_perm_load): Likewise.
15829 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
15830 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
15831 (vect_gen_perm_mask_checked): Likewise.
15832 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
15833 (vect_gen_perm_mask_checked): Likewise.
15834 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
15835 (vectorizable_store): Likewise.
15836 (vectorizable_load): Likewise.
15837 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
15838 (vectorizable_bswap): Likewise.
15840 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15841 Alan Hayward <alan.hayward@arm.com>
15842 David Sherwood <david.sherwood@arm.com>
15844 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
15845 * tree.c (build_vector): Likewise.
15846 (build_vector_from_ctor): Update accordingly.
15847 (build_vector_from_val): Likewise.
15848 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
15849 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
15850 * tree-vect-generic.c (add_rshift): Likewise.
15851 (expand_vector_divmod): Likewise.
15852 (optimize_vector_constructor): Likewise.
15853 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
15854 (vect_transform_slp_perm_load): Likewise.
15855 (vect_schedule_slp_instance): Likewise.
15856 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
15857 (vectorizable_call): Likewise.
15858 (vect_gen_perm_mask_any): Likewise. Add elements in order.
15859 * expmed.c (make_tree): Likewise.
15860 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
15861 a vector passed to build_vector.
15862 (fold_convert_const): Likewise.
15863 (exact_inverse): Likewise.
15864 (fold_ternary_loc): Likewise.
15865 (fold_relational_const): Likewise.
15866 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
15867 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
15868 (const_unop): Likewise. Store the reduction accumulator in a
15869 variable rather than an array.
15870 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
15871 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
15872 the new vector, rather than constructing it after the input arrays.
15873 (native_interpret_vector): Use auto_vec<tree> when building
15874 a vector passed to build_vector. Add elements in order.
15875 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
15876 auto_vec<tree> when building a vector passed to build_vector.
15877 (vect_create_epilog_for_reduction): Likewise.
15878 (vectorizable_induction): Likewise.
15879 (get_initial_def_for_reduction): Likewise. Fix indentation of
15881 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
15883 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
15884 passed to build_vector.
15886 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
15887 Alan Hayward <alan.hayward@arm.com>
15888 David Sherwood <david.sherwood@arm.com>
15890 * tree-core.h (tree_base::u): Add an "nelts" field.
15891 (tree_vector): Use VECTOR_CST_NELTS as the length.
15892 * tree.c (tree_size): Likewise.
15893 (make_vector): Initialize VECTOR_CST_NELTS.
15894 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
15895 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
15896 TYPE_VECTOR_SUBPARTS.
15897 * expr.c (const_vector_mask_from_tree): Consistently use "units"
15898 as the number of units, setting it from VECTOR_CST_NELTS.
15899 (const_vector_from_tree): Likewise.
15900 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
15901 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
15902 (fold_negate_expr_1): Likewise.
15903 (fold_convert_const): Likewise.
15904 (const_binop): Likewise. Differentiate the number of output and
15906 (const_unop): Likewise.
15907 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
15908 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
15909 in cases that did the opposite.
15911 2017-09-14 Richard Biener <rguenther@suse.de>
15913 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
15916 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
15918 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
15920 2017-09-14 Jakub Jelinek <jakub@redhat.com>
15923 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
15924 if and where to split a bb, except for splitting before debug insn
15925 sequences followed by non-label real insn. Delete debug insns
15926 in between basic blocks.
15928 * combine.c (make_compound_operation_int): Formatting fixes.
15930 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
15931 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15932 * config/netbsd.h (LINK_EH_SPEC): Likewise.
15933 * config/sol2.h (LINK_EH_SPEC): Likewise.
15934 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15935 * config/s390/linux.h (LINK_SPEC): Likewise.
15936 * config/freebsd.h (LINK_EH_SPEC): Likewise.
15937 * config/openbsd.h (LINK_EH_SPEC): Likewise.
15938 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15939 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
15940 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
15941 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
15942 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
15943 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
15945 2017-09-13 Jakub Jelinek <jakub@redhat.com>
15947 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
15949 (ENDFILE_LINUX_SPEC): Likewise.
15950 (LINK_EH_SPEC): Likewise.
15951 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
15952 (LINK_OS_LINUX_SPEC32): Likewise.
15953 (LINK_OS_LINUX_SPEC64): Likewise.
15954 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
15955 (LINK_OS_LINUX_SPEC): Likewise.
15957 2017-09-13 Martin Liska <mliska@suse.cz>
15959 PR middle-end/82154
15960 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
15961 CASE_HIGH is NULL_TREE.
15963 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
15964 Alan Hayward <alan.hayward@arm.com>
15965 David Sherwood <david.sherwood@arm.com>
15967 * target.def (secondary_memory_needed): New hook.
15968 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
15969 instead of SECONDARY_MEMORY_NEEDED.
15970 (secondary_memory_needed_mode): Likewise.
15971 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
15972 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
15973 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
15974 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
15975 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
15976 * doc/tm.texi: Regenerate.
15977 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
15978 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
15979 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15980 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
15981 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
15982 * config/i386/i386.c (inline_secondary_memory_needed): Put the
15983 mode argument first and change the reg_class arguments to reg_class_t.
15984 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
15985 Make static. Update the call to inline_secondary_memory_needed.
15986 (ix86_register_move_cost): Update the call to
15987 inline_secondary_memory_needed.
15988 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15989 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
15991 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
15992 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
15994 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
15995 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
15996 * config/mips/mips.c (mips_secondary_memory_needed): Make static
15997 and match hook interface. Add comment from mips.h.
15998 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
15999 * config/mmix/mmix.md (truncdfsf2): Refer to
16000 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
16002 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
16003 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
16004 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16005 (pa_secondary_memory_needed): New function.
16006 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
16007 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
16008 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16009 (pdp11_secondary_memory_needed): Make static and match hook interface.
16010 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
16011 * config/powerpcspe/powerpcspe-protos.h
16012 (rs6000_secondary_memory_needed_ptr): Delete.
16013 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
16015 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16016 (rs6000_option_override_internal): Assign to
16017 targetm.secondary_memory_needed rather than
16018 rs6000_secondary_memory_needed_ptr.
16019 (rs6000_secondary_memory_needed): Match hook interface.
16020 (rs6000_debug_secondary_memory_needed): Likewise.
16021 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
16022 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
16023 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
16024 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16025 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
16026 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
16028 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
16029 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16030 (rs6000_option_override_internal): Assign to
16031 targetm.secondary_memory_needed rather than
16032 rs6000_secondary_memory_needed_ptr.
16033 (rs6000_secondary_memory_needed): Match hook interface.
16034 (rs6000_debug_secondary_memory_needed): Likewise.
16035 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
16036 * config/s390/s390.c (s390_secondary_memory_needed): New function.
16037 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16038 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
16039 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
16040 (sparc_secondary_memory_needed): New function.
16041 * lra-constraints.c (check_and_process_move): Refer to
16042 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
16044 (curr_insn_transform): Likewise.
16045 (process_alt_operands): Use targetm.secondary_memory_needed
16046 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16047 (check_secondary_memory_needed_p): Likewise.
16048 (choose_split_class): Likewise.
16049 * reload.c: Unconditionally include code that was previously
16050 conditional on SECONDARY_MEMORY_NEEDED.
16051 (push_secondary_reload): Use targetm.secondary_memory_needed
16052 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16053 (push_reload): Likewise.
16054 * reload1.c: Unconditionally include code that was previously
16055 conditional on SECONDARY_MEMORY_NEEDED.
16056 (choose_reload_regs): Use targetm.secondary_memory_needed
16057 instead of TARGET_SECONDARY_MEMORY_NEEDED.
16058 (gen_reload): Likewise.
16059 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
16061 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
16062 Alan Hayward <alan.hayward@arm.com>
16063 David Sherwood <david.sherwood@arm.com>
16065 * target.def (secondary_memory_needed_mode): New hook:
16066 * targhooks.c (default_secondary_memory_needed_mode): Declare.
16067 * targhooks.h (default_secondary_memory_needed_mode): New function.
16068 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
16069 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
16070 * doc/tm.texi: Regenerate.
16071 * lra-constraints.c (check_and_process_move): Use
16072 targetm.secondary_memory_needed_mode instead of
16073 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
16074 (curr_insn_transform): Likewise.
16075 * reload.c (get_secondary_mem): Likewise.
16076 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16077 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
16079 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16080 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16081 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
16082 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16083 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
16085 * config/powerpcspe/powerpcspe-protos.h
16086 (rs6000_secondary_memory_needed_mode): Delete.
16087 * config/powerpcspe/powerpcspe.c
16088 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16089 (rs6000_secondary_memory_needed_mode): Make static.
16090 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16091 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
16093 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16095 (rs6000_secondary_memory_needed_mode): Make static.
16096 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16097 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
16098 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
16099 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
16100 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
16102 (sparc_secondary_memory_needed_mode): New function.
16103 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
16105 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
16107 * config/aarch64/constraints.md (Umq): New constraint.
16108 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
16110 (mov<mode>): Update condition.
16112 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16114 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
16115 when bitposition is the same.
16117 2017-09-13 Richard Biener <rguenther@suse.de>
16119 * dwarf2out.c (output_die_symbol): Remove.
16120 (output_die): Do not output a DIEs symbol.
16122 2017-09-13 Richard Biener <rguenther@suse.de>
16124 PR middle-end/82128
16125 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
16126 default-def to avoid breaking iterator update with the weird
16127 interaction with cgraph_update_edges_for_call_stmt_node.
16129 2017-09-13 Richard Biener <rguenther@suse.de>
16131 * tree-cfg.c (verify_gimple_assign_binary): Add verification
16132 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
16133 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
16134 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
16136 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
16138 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
16139 Disable pc relative literal load irrespective of
16140 TARGET_FIX_ERR_A53_84341 for default.
16142 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
16144 * config/sparc/sparc.c (output_return): Output the source location of
16145 the insn in the delay slot, if any.
16146 (output_sibcall): Likewise.
16148 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
16151 * common.opt (-static-pie): New alias.
16152 (shared): Negate static-pie.
16153 (-no-pie): Update help text.
16155 (static-pie): New option.
16156 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
16157 -static-pie support.
16158 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
16159 (LINK_EH_SPEC): Likewise.
16160 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
16161 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16162 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
16163 * gcc.c (LINK_COMMAND_SPEC): Likewise.
16164 (init_gcc_specs): Likewise.
16165 (init_spec): Likewise.
16166 (display_help): Update help message for -pie.
16167 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
16170 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
16172 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
16173 (movdi_aarch64): Likewise.
16174 (movti_aarch64): Likewise.
16176 2017-09-12 Simon Wright <simon@pushface.org>
16179 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
16180 calculation of the minor version, always output as 0.
16182 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16185 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
16186 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
16187 on it early, rather than manual conversion late. For
16188 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
16189 instead of performing manual conversion.
16191 2017-09-12 Carl Love <cel@us.ibm.com>
16193 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
16194 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
16196 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16197 VMULOSW): Add definitions.
16198 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16199 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16200 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16201 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16202 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16204 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16206 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
16208 (movti_aarch64): Likewise.
16209 (movdf_aarch64): Likewise.
16210 (movtf_aarch64): Likewise.
16211 (load_pairdi): Likewise.
16212 (store_pairdi): Likewise.
16213 (load_pairdf): Likewise.
16214 (store_pairdf): Likewise.
16215 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
16216 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
16217 (ldr_got_small_<mode>): Likewise.
16218 (ldr_got_small_28k_<mode>): Likewise.
16219 (ldr_got_tiny): Likewise.
16220 * config/aarch64/iterators.md (ldst_sz): New.
16221 (ldpstp_sz): Likewise.
16222 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
16224 (thunderx_load): Split load_8 to load_16.
16225 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
16227 (thunderx2t99_storepair_basic): Split store_8 to store_16.
16228 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
16229 (xgene1_store_pair): Split store_8 to store_16.
16230 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
16231 (falkor_st_0_st_sd): Split store_8 to store_16.
16233 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
16235 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
16236 and store1/2/3/4 to store_4/8/12/16.
16237 * config/aarch64/aarch64.md: Update for rename.
16238 * config/arm/arm.md: Likewise.: Likewise.
16239 * config/arm/arm.c: Likewise.
16240 * config/arm/thumb1.md: Likewise.
16241 * config/arm/thumb2.md: Likewise.
16242 * config/arm/vfp.md: Likewise.
16243 * config/arm/arm-generic.md: Likewise.
16244 * config/arm/arm1020e.md: Likewise.
16245 * config/arm/arm1026ejs.md: Likewise.
16246 * config/arm/arm1136jfs.md: Likewise.
16247 * config/arm/arm926ejs.md: Likewise.
16248 * config/arm/cortex-a15.md: Likewise.
16249 * config/arm/cortex-a17.md: Likewise.
16250 * config/arm/cortex-a5.md: Likewise.
16251 * config/arm/cortex-a53.md: Likewise.
16252 * config/arm/cortex-a57.md: Likewise.
16253 * config/arm/cortex-a7.md: Likewise.
16254 * config/arm/cortex-a8.md: Likewise.
16255 * config/arm/cortex-a9.md: Likewise.
16256 * config/arm/cortex-m4.md: Likewise.
16257 * config/arm/cortex-m7.md: Likewise.
16258 * config/arm/cortex-r4.md: Likewise.
16259 * config/arm/exynos-m1.md: Likewise.
16260 * config/arm/fa526.md: Likewise.
16261 * config/arm/fa606te.md: Likewise.
16262 * config/arm/fa626te.md: Likewise.
16263 * config/arm/fa726te.md: Likewise.
16264 * config/arm/fmp626.md: Likewise.
16265 * config/arm/iwmmxt.md: Likewise.
16266 * config/arm/ldmstm.md: Likewise.
16267 * config/arm/marvell-pj4.md: Likewise.
16268 * config/arm/xgene1.md: Likewise.
16269 * config/aarch64/thunderx.md: Likewise.
16270 * config/aarch64/thunderx2t99.md: Likewise.
16271 * config/aarch64/falkor.md: Likewise.
16273 2017-09-12 Martin Liska <mliska@suse.cz>
16275 * attribs.c (private_lookup_attribute): New function.
16276 * attribs.h (private_lookup_attribute): Declared here.
16277 (lookup_attribute): Called from this place.
16279 2017-09-12 Richard Biener <rguenther@suse.de>
16281 PR tree-optimization/82157
16282 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
16283 stmts with side-effects.
16285 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16286 Alan Hayward <alan.hayward@arm.com>
16287 David Sherwood <david.sherwood@arm.com>
16289 * target.def (hard_regno_nregs): New hook.
16290 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
16291 * targhooks.h (default_hard_regno_nregs): Declare.
16292 * targhooks.c (default_hard_regno_nregs): New function.
16293 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
16294 (TARGET_HARD_REGNO_NREGS): ...this hook.
16295 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
16296 (CLASS_MAX_NREGS): Likewise.
16297 * doc/tm.texi: Regenerate.
16298 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
16299 instead of HARD_REGNO_NREGS.
16300 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
16301 HARD_REGNO_NREGS in the comment.
16302 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
16303 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
16304 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
16305 Return an unsigned int.
16306 (TARGET_HARD_REGNO_NREGS): Redefine.
16307 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
16308 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
16309 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16310 (arc_hard_regno_nregs): New function.
16311 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
16312 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
16313 (arm_hard_regno_nregs): New function.
16314 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
16315 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
16316 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
16317 (TARGET_HARD_REGNO_NREGS): Redefine.
16318 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
16319 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
16320 (HARD_REGNO_NREGS): Delete.
16321 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
16322 (cr16_hard_regno_nregs): New function.
16323 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
16324 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
16325 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
16326 (cris_hard_regno_nregs): New function.
16327 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
16328 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
16329 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16330 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
16331 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
16332 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
16333 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
16334 (frv_hard_regno_nregs): Make static. Take and return an
16336 (frv_class_max_nregs): Remove outdated copy of documentation.
16337 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
16338 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
16339 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
16340 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
16341 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
16342 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
16343 (TARGET_HARD_REGNO_NREGS): Redefine.
16344 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
16345 (CLASS_MAX_NREGS): Update comment.
16346 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
16347 (ia64_hard_regno_nregs): New function.
16348 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
16349 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
16350 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
16351 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
16352 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
16354 (m32c_hard_regno_nregs): Likewise. Make static.
16355 (TARGET_HARD_REGNO_NREGS): Redefine.
16356 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
16357 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
16358 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
16359 (m68k_hard_regno_nregs): New function.
16360 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
16361 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
16362 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
16363 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
16364 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
16365 Take and return an unsigned int.
16366 (TARGET_HARD_REGNO_NREGS): Redefine.
16367 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
16368 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
16369 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
16370 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
16371 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
16372 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
16373 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
16374 (msp430_hard_regno_nregs): Make static. Take and return an
16376 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
16377 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
16378 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
16379 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
16380 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
16381 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
16382 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
16383 (TARGET_HARD_REGNO_NREGS): Redefine.
16384 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
16385 (PA_HARD_REGNO_NREGS): ...this.
16386 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
16387 (PA_HARD_REGNO_NREGS): ...this.
16388 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16389 (pa_hard_regno_nregs): New function.
16390 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
16391 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
16392 (pdp11_hard_regno_nregs): New function.
16393 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
16394 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
16395 (rs6000_hard_regno_nregs_hook): New function.
16396 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
16397 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
16398 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
16399 Take and return an unsigned int. Move earlier in file.
16400 (TARGET_HARD_REGNO_NREGS): Redefine.
16401 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
16402 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
16403 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
16404 (rl78_hard_regno_nregs): Make static. Take and return an
16406 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
16407 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
16408 (rs6000_hard_regno_nregs_hook): New function.
16409 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
16410 * config/rx/rx.c (rx_hard_regno_nregs): New function.
16411 (TARGET_HARD_REGNO_NREGS): Redefine.
16412 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
16413 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
16414 instead of HARD_REGNO_NREGS.
16415 (s390_hard_regno_nregs): New function.
16416 (s390_hard_regno_mode_ok): Add comment from s390.h.
16417 (TARGET_HARD_REGNO_NREGS): Redefine.
16418 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
16419 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
16420 (sh_hard_regno_nregs): New function.
16421 (sh_pass_in_reg_p): Use it.
16422 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
16423 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
16424 (sparc_hard_regno_nregs): New function.
16425 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
16426 * config/spu/spu.c (spu_hard_regno_nregs): New function.
16427 (spu_function_arg_advance): Use it, supplying a valid register number.
16428 (TARGET_HARD_REGNO_NREGS): Redefine.
16429 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
16430 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
16431 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
16432 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
16433 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
16434 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
16435 (CLASS_MAX_NREGS): Remove copy of old documentation.
16436 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
16437 (visium_hard_regno_nregs): New function.
16438 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
16439 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
16440 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
16441 xtensa_hard_regno_nregs): New function.
16442 * system.h (HARD_REGNO_NREGS): Poison.
16444 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16446 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
16447 hard_regno_nregs instead of HARD_REGNO_NREGS.
16448 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
16449 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
16450 (c6x_expand_epilogue): Likewise.
16451 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
16452 (frv_read_iacc_argument): Likewise.
16453 * config/sh/sh.c: Include regs.h.
16454 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16455 (regs_used): Likewise.
16456 (output_stack_adjust): Likewise.
16457 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
16458 * expmed.c: Include regs.h.
16459 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
16460 * ree.c: Include regs.h.
16461 (combine_reaching_defs): Use hard_regno_nregs instead of
16463 (add_removable_extension): Likewise.
16465 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16467 * regs.h (hard_regno_nregs): Turn into a function.
16468 (end_hard_regno): Update accordingly.
16469 * caller-save.c (setup_save_areas): Likewise.
16470 (save_call_clobbered_regs): Likewise.
16471 (replace_reg_with_saved_mem): Likewise.
16472 (insert_restore): Likewise.
16473 (insert_save): Likewise.
16474 * combine.c (can_change_dest_mode): Likewise.
16475 (move_deaths): Likewise.
16476 (distribute_notes): Likewise.
16477 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
16478 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
16479 (rs6000_split_multireg_move): Likewise.
16480 (rs6000_register_move_cost): Likewise.
16481 (rs6000_memory_move_cost): Likewise.
16482 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
16483 (rs6000_split_multireg_move): Likewise.
16484 (rs6000_register_move_cost): Likewise.
16485 (rs6000_memory_move_cost): Likewise.
16486 * cselib.c (cselib_reset_table): Likewise.
16487 (cselib_lookup_1): Likewise.
16488 * emit-rtl.c (set_mode_and_regno): Likewise.
16489 * function.c (aggregate_value_p): Likewise.
16490 * ira-color.c (setup_profitable_hard_regs): Likewise.
16491 (check_hard_reg_p): Likewise.
16492 (calculate_saved_nregs): Likewise.
16493 (assign_hard_reg): Likewise.
16494 (improve_allocation): Likewise.
16495 (calculate_spill_cost): Likewise.
16496 * ira-emit.c (modify_move_list): Likewise.
16497 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
16498 (ira_hard_reg_in_set_p): Likewise.
16499 * ira.c (setup_reg_mode_hard_regset): Likewise.
16500 (clarify_prohibited_class_mode_regs): Likewise.
16501 (check_allocation): Likewise.
16502 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16503 (lra_setup_reg_renumber): Likewise.
16504 (setup_try_hard_regno_pseudos): Likewise.
16505 (spill_for): Likewise.
16506 (assign_hard_regno): Likewise.
16507 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
16508 * lra-constraints.c (in_class_p): Likewise.
16509 (lra_constraint_offset): Likewise.
16510 (simplify_operand_subreg): Likewise.
16511 (lra_constraints): Likewise.
16512 (split_reg): Likewise.
16513 (split_if_necessary): Likewise.
16514 (invariant_p): Likewise.
16515 (inherit_in_ebb): Likewise.
16516 * lra-lives.c (process_bb_lives): Likewise.
16517 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
16518 (get_hard_regs): Likewise.
16519 (do_remat): Likewise.
16520 * lra-spills.c (assign_spill_hard_regs): Likewise.
16521 * mode-switching.c (create_pre_exit): Likewise.
16522 * postreload.c (reload_combine_recognize_pattern): Likewise.
16523 * recog.c (peep2_find_free_register): Likewise.
16524 * regcprop.c (kill_value_regno): Likewise.
16525 (set_value_regno): Likewise.
16526 (copy_value): Likewise.
16527 (maybe_mode_change): Likewise.
16528 (find_oldest_value_reg): Likewise.
16529 (copyprop_hardreg_forward_1): Likewise.
16530 * regrename.c (check_new_reg_p): Likewise.
16531 (regrename_do_replace): Likewise.
16532 * reload.c (push_reload): Likewise.
16533 (combine_reloads): Likewise.
16534 (find_dummy_reload): Likewise.
16535 (operands_match_p): Likewise.
16536 (find_reloads): Likewise.
16537 (find_equiv_reg): Likewise.
16538 (reload_adjust_reg_for_mode): Likewise.
16539 * reload1.c (count_pseudo): Likewise.
16540 (count_spilled_pseudo): Likewise.
16541 (find_reg): Likewise.
16542 (clear_reload_reg_in_use): Likewise.
16543 (free_for_value_p): Likewise.
16544 (allocate_reload_reg): Likewise.
16545 (choose_reload_regs): Likewise.
16546 (reload_adjust_reg_for_temp): Likewise.
16547 (emit_reload_insns): Likewise.
16548 (delete_output_reload): Likewise.
16549 * rtlanal.c (subreg_get_info): Likewise.
16550 * sched-deps.c (sched_analyze_reg): Likewise.
16551 * sel-sched.c (init_regs_for_mode): Likewise.
16552 (mark_unavailable_hard_regs): Likewise.
16553 (choose_best_reg_1): Likewise.
16554 (verify_target_availability): Likewise.
16555 * valtrack.c (dead_debug_insert_temp): Likewise.
16556 * var-tracking.c (track_loc_p): Likewise.
16557 (emit_note_insn_var_location): Likewise.
16558 * varasm.c (make_decl_rtl): Likewise.
16559 * reginfo.c (choose_hard_reg_mode): Likewise.
16560 (init_reg_modes_target): Refer directly to
16561 this_target_regs->x_hard_regno_nregs.
16563 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16565 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
16566 instead of hard_regno_nregs.
16568 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16570 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
16571 end_hard_regno instead of hard_regno_nregs.
16572 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
16573 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
16574 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
16575 * ira-color.c (improve_allocation): Likewise.
16576 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16577 * lra-lives.c (mark_regno_live): Likewise.
16578 (mark_regno_dead): Likewise.
16579 * lra-remat.c (operand_to_remat): Likewise.
16580 * lra.c (collect_non_operand_hard_regs): Likewise.
16581 * postreload.c (reload_combine_note_store): Likewise.
16582 (move2add_valid_value_p): Likewise.
16583 * reload.c (regno_clobbered_p): Likewise.
16585 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16587 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
16589 * config/v850/v850.c (v850_reorg): Likewise.
16590 * reload.c (refers_to_regno_for_reload_p): Likewise.
16591 (find_equiv_reg): Likewise.
16592 * reload1.c (reload_reg_reaches_end_p): Likewise.
16594 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16596 * caller-save.c (add_used_regs): Use REG_NREGS instead of
16598 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
16599 * config/arm/arm.c (output_move_neon): Likewise.
16600 (arm_attr_length_move_neon): Likewise.
16601 (neon_split_vcombine): Likewise.
16602 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
16603 (c6x_mark_reg_written): Likewise.
16604 (c6x_dwarf_register_span): Likewise.
16605 * config/i386/i386.c (ix86_save_reg): Likewise.
16606 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
16607 (rws_access_reg): Likewise.
16608 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
16609 * mode-switching.c (create_pre_exit): Likewise.
16610 * ree.c (combine_reaching_defs): Likewise.
16611 (add_removable_extension): Likewise.
16612 * regcprop.c (find_oldest_value_reg): Likewise.
16613 (copyprop_hardreg_forward_1): Likewise.
16614 * reload.c (reload_inner_reg_of_subreg): Likewise.
16615 (push_reload): Likewise.
16616 (combine_reloads): Likewise.
16617 (find_dummy_reload): Likewise.
16618 (reload_adjust_reg_for_mode): Likewise.
16619 * reload1.c (find_reload_regs): Likewise.
16620 (forget_old_reloads_1): Likewise.
16621 (reload_reg_free_for_value_p): Likewise.
16622 (reload_adjust_reg_for_temp): Likewise.
16623 (emit_reload_insns): Likewise.
16624 (delete_output_reload): Likewise.
16625 * sel-sched.c (choose_best_reg_1): Likewise.
16626 (choose_best_pseudo_reg): Likewise.
16628 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16629 Alan Hayward <alan.hayward@arm.com>
16630 David Sherwood <david.sherwood@arm.com>
16632 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
16633 * target.def (slow_unaligned_access): New hook.
16634 * targhooks.h (default_slow_unaligned_access): Declare.
16635 * targhooks.c (default_slow_unaligned_access): New function.
16636 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
16637 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
16638 * doc/tm.texi: Regenerate.
16639 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
16640 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
16641 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
16643 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
16644 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
16646 (rs6000_slow_unaligned_access): New function.
16647 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16648 (expand_block_compare): Likewise.
16649 (expand_strn_compare): Likewise.
16650 (rs6000_rtx_costs): Likewise.
16651 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
16652 (riscv_slow_unaligned_access): Likewise.
16653 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
16654 (riscv_slow_unaligned_access_p): ...this and make static.
16655 (riscv_option_override): Update accordingly.
16656 (riscv_slow_unaligned_access): New function.
16657 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16658 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
16659 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
16660 (rs6000_slow_unaligned_access): New function.
16661 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
16662 (rs6000_rtx_costs): Likewise.
16663 * config/rs6000/rs6000-string.c (expand_block_compare)
16664 (expand_strn_compare): Use targetm.slow_unaligned_access instead
16665 of SLOW_UNALIGNED_ACCESS.
16666 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
16667 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
16668 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
16669 of SLOW_UNALIGNED_ACCESS.
16670 * expmed.c (simple_mem_bitfield_p): Likewise.
16671 * expr.c (alignment_for_piecewise_move): Likewise.
16672 (emit_group_load_1): Likewise.
16673 (emit_group_store): Likewise.
16674 (copy_blkmode_from_reg): Likewise.
16675 (emit_push_insn): Likewise.
16676 (expand_assignment): Likewise.
16677 (store_field): Likewise.
16678 (expand_expr_real_1): Likewise.
16679 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
16680 * lra-constraints.c (simplify_operand_subreg): Likewise.
16681 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
16682 * gimple-ssa-store-merging.c: Likewise in block comment at start
16684 * tree-ssa-strlen.c: Include target.h.
16685 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
16686 of SLOW_UNALIGNED_ACCESS.
16687 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
16689 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
16691 PR rtl-optimization/82185
16692 * expmed.c (emit_store_flag_int): Only test tem if it has been
16695 2017-09-12 Richard Biener <rguenther@suse.de>
16697 PR middle-end/82149
16698 * match.pd ((FTYPE) N CMP CST): Fix typo.
16700 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
16702 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
16704 (mips_near_type_p): Add 'short_call' attribute as a synonym
16706 * doc/extend.texi (short_call): Document new function attribute.
16708 2017-09-12 Jakub Jelinek <jakub@redhat.com>
16711 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
16712 assertion check that in the condition.
16713 (get_atomic_generic_size): Likewise. Before testing if parameter
16714 has pointer type, if it has array type, call for C++
16715 default_conversion to perform array-to-pointer conversion.
16717 2017-09-12 Richard Biener <rguenther@suse.de>
16719 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
16720 for operations we cannot scalarize.
16722 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
16724 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
16725 vectors heap vectors. Clean up comments.
16726 Make visited_bbs a reference.
16727 (profitable_jump_thread_path): Make GC
16728 vectors heap vectors. Clean up comments.
16730 (convert_and_register_jump_thread_path): Make GC vectors heap
16732 (check_subpath_and_update_thread_path): Same. Clean up comments.
16733 Make visited_bbs a reference.
16734 (handle_phi): Abstract common code to to
16735 register_jump_thread_path_if_profitable.
16736 Rename VAR_BB to DEF_BB.
16738 Make GC vectors heap vectors.
16739 Make visited_bbs a reference.
16740 (handle_assignment): Same.
16741 (register_jump_thread_path_if_profitable): New.
16742 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
16744 Make GC vectors heap vectors. Clean up comments.
16745 Make visited_bbs a reference.
16746 (find_jump_threads_backwards): Make visited_bbs live in the stack.
16747 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
16750 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
16753 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
16754 words of E_DImode object are reachable by xtensa_uimm8x4 access.
16756 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
16758 Revert r251800 and r251799.
16760 2017-09-11 Martin Jambor <mjambor@suse.cz>
16763 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
16764 arguments in advance.
16765 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
16766 use it to find predecessor edges.
16767 (naive_outof_ssa): Collect vector of predecessors.
16769 2017-09-08 Jason Merrill <jason@redhat.com>
16771 PR c++/70029 - ICE with ref-qualifier and -flto
16772 * langhooks.h (struct lang_hooks_for_types): Add
16773 copy_lang_qualifiers.
16774 * attribs.c (build_type_attribute_qual_variant): Use it.
16775 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
16777 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
16778 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
16780 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
16783 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
16784 (*mulsi3_sp64): New instruction.
16785 (mulsi3): New expander.
16787 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
16789 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
16791 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16793 * sancov.c: Include memmodel.h.
16795 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
16798 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
16801 2017-09-07 Carl Love <cel@us.ibm.com>
16803 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
16804 the sldi instruction.
16806 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
16808 * sancov.c: Include tm_p.h.
16810 2017-09-07 Jakub Jelinek <jakub@redhat.com>
16813 * output.h (switch_to_other_text_partition): New declaration.
16814 * varasm.c (switch_to_other_text_partition): New function.
16815 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
16816 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
16817 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
16818 to the other text partition before emitting LCL label and switch back
16819 after emitting the word after it.
16821 2017-09-07 Richard Biener <rguenther@suse.de>
16823 * passes.def (pass_split_crit_edges): Remove instance before PRE.
16824 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
16825 critical edges here, after loop init.
16826 (pass_data_pre): Remove PROP_no_crit_edges flags.
16827 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
16828 for valueization of call args to avoid leaking VN_TOP.
16829 (visit_use): Assert we do not visit default defs.
16830 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
16831 Use error_mark_node to more easily detect leaking VN_TOP.
16832 All default-defs are varying, not VN_TOP. Mark them visited.
16833 (run_scc_vn): Make code match comment.
16835 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16837 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
16838 OPTION_MASK_FLOAT128_KEYWORD.
16839 (POWERPC_MASKS): Likewise.
16840 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
16841 support for the -mfloat128-type option, and make -mfloat128
16842 default on PowerPC Linux systems. Define or undefine
16843 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
16844 Define __float128 to be __ieee128 if IEEE 128-bit support is
16845 enabled, or undefine it.
16846 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
16847 Delete defining __FLOAT128_TYPE__.
16848 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
16849 -mfloat128-type option and make -mfloat128 default on PowerPC
16851 (TARGET_FLOAT128_TYPE): Likewise.
16852 (-mfloat128-type): Likewise.
16853 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16854 Delete the -mfloat128-type option and make -mfloat128 default on
16855 PowerPC Linux systems. Always use __ieee128 to be the keyword for
16856 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
16857 128-bit floating point is enabled. Change tests from using
16858 -mfloat128-type to -mfloat128.
16859 (rs6000_mangle_type): Use the correct mangling for the __float128
16860 type even if normal long double is restricted to 64-bits.
16861 (floatn_mode): Enable the _Float128 type by default on VSX Linux
16863 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
16864 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
16865 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
16867 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
16868 documentation for -mfloat128.
16870 2017-09-06 Olivier Hainque <hainque@adacore.com>
16872 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
16874 2017-09-06 Wish Wu <wishwu007@gmail.com>
16875 Jakub Jelinek <jakub@redhat.com>
16877 * asan.c (initialize_sanitizer_builtins): Add
16878 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
16879 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
16880 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
16881 BT_FN_VOID_UINT64_PTR variables.
16882 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
16883 (BT_FN_VOID_UINT16_UINT16): Likewise.
16884 (BT_FN_VOID_UINT32_UINT32): Likewise.
16885 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
16886 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
16887 (BT_FN_VOID_UINT64_PTR): Likewise.
16888 * common.opt (flag_sanitize_coverage): New variable.
16889 (fsanitize-coverage=trace-pc): Remove.
16890 (fsanitize-coverage=): Add.
16891 * flag-types.h (enum sanitize_coverage_code): New enum.
16892 * fold-const.c (fold_range_test): Disable non-short-circuit
16893 optimization if flag_sanitize_coverage.
16894 (fold_truth_andor): Likewise.
16895 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
16896 * opts.c (COVERAGE_SANITIZER_OPT): Define.
16897 (coverage_sanitizer_opts): New array.
16898 (get_closest_sanitizer_option): Add OPTS argument, handle also
16899 OPT_fsanitize_coverage_.
16900 (parse_sanitizer_options): Adjusted to also handle
16901 OPT_fsanitize_coverage_.
16902 (common_handle_option): Add OPT_fsanitize_coverage_.
16903 * sancov.c (instrument_comparison, instrument_switch): New function.
16904 (sancov_pass): Add trace-cmp support.
16905 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
16906 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
16907 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
16908 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
16909 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
16910 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
16911 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
16912 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
16913 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
16914 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
16916 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16918 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
16919 error. Only quit immediately if parsing is complete.
16920 (BEGIN): Initialize fatal_err and parse_done.
16921 (begin fpu, end fpu): Check number of arguments.
16922 (begin arch, end arch): Likewise.
16923 (begin cpu, end cpu): Likewise.
16924 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
16925 (optalias): Likewise.
16927 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
16929 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
16930 * config/arm/arm-isa.h: Delete. Move definitions to ...
16931 * arm-cpus.in: ... here. Use new feature and fgroup values.
16932 * config/arm/arm.c (arm_option_override): Use lower case for feature
16934 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
16935 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
16936 * config/arm/parsecpu.awk (END): Add new command 'isa'.
16938 (print_isa_bits_for): New function.
16939 (gen_isa): New function.
16940 (gen_comm_data): Use print_isa_bits_for.
16941 (define feature): New keyword.
16942 (define fgroup): New keyword.
16943 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
16944 (arm-isa.h): Add rule to generate file.
16945 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
16946 case for feature bit names.
16948 2017-09-06 Richard Biener <rguenther@suse.de>
16950 * tree-ssa-pre.c (NECESSARY): Remove.
16951 (create_expression_by_pieces): Do not touch pass-local flags.
16952 (insert_into_preds_of_block): Likewise.
16953 (do_pre_regular_insertion): Likewise.
16954 (eliminate_insert): Likewise.
16955 (eliminate_dom_walker::before_dom_children): Likewise.
16956 (fini_eliminate): Do not look at inserted_exprs.
16957 (mark_operand_necessary): Remove.
16958 (remove_dead_inserted_code): Replace with simple work-list
16959 algorithm based on inserted_exprs and SSA uses.
16960 (pass_pre::execute): Re-order fini_eliminate and
16961 remove_dead_inserted_code.
16963 2017-09-06 Olivier Hainque <hainque@adacore.com>
16965 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
16966 for VxWorks 7. Adjust surrounding comments.
16968 2017-09-06 Richard Biener <rguenther@suse.de>
16970 * gimple-ssa-strength-reduction.c
16971 (find_candidates_dom_walker::before_dom_children): Also allow
16974 2017-09-06 Richard Biener <rguenther@suse.de>
16976 PR tree-optimization/82108
16977 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
16978 for gap in the non-permutation SLP case.
16980 2017-09-06 Martin Jambor <mjambor@suse.cz>
16982 PR tree-optimization/82078
16983 * tree-sra.c (sort_and_splice_var_accesses): Move call to
16984 add_access_to_work_queue...
16985 (build_accesses_from_assign): ...here.
16986 (propagate_all_subaccesses): Make sure racc is the group
16987 representative, if there is one.
16989 2017-09-06 Jakub Jelinek <jakub@redhat.com>
16991 PR middle-end/82095
16992 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
16995 2017-09-06 Richard Biener <rguenther@suse.de>
16997 * gimple-ssa-strength-reduction.c
16998 (find_candidates_dom_walker::before_doom_children): Use a
16999 type and not a mode check.
17001 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
17004 * config/arm/predicates.md (arm_general_adddi_operand): Create new
17006 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
17008 2017-09-05 Jeff Law <law@redhat.com>
17010 PR tree-optimization/64910
17011 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
17012 cases where we have 3 or more operands.
17014 2017-09-05 Jakub Jelinek <jakub@redhat.com>
17016 PR middle-end/81768
17017 * omp-low.c (lower_omp_for): Recompute tree invariant if
17018 gimple_omp_for_initial/final is ADDR_EXPR.
17020 PR middle-end/81768
17021 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
17022 into gimple val before gimplification fo the COND_EXPR.
17024 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
17026 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
17027 REGION_COPY argument.
17028 (thread_through_all_blocks): Remove unused argument to
17029 duplicate_thread_path.
17031 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17032 Alan Hayward <alan.hayward@arm.com>
17033 David Sherwood <david.sherwood@arm.com>
17035 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
17036 Take a scalar_mode rather than a machine_mode.
17037 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17038 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
17039 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17040 (aarch64_gen_adjusted_ldpstp): Likewise.
17041 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
17043 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17044 Alan Hayward <alan.hayward@arm.com>
17045 David Sherwood <david.sherwood@arm.com>
17047 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
17048 Take a scalar_int_mode instead of a machine_mode.
17049 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
17050 (aarch64_output_scalar_simd_mov_immediate): Likewise.
17051 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
17052 (aarch64_simd_attr_length_rglist): Delete.
17053 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
17054 a scalar_int_mode instead of a machine_mode.
17055 (aarch64_add_offset): Likewise.
17056 (aarch64_internal_mov_immediate): Likewise
17057 (aarch64_add_constant_internal): Likewise.
17058 (aarch64_add_constant): Likewise.
17059 (aarch64_movw_imm): Likewise.
17060 (aarch64_rtx_arith_op_extract_p): Likewise.
17061 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
17062 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
17063 Remove assert that the mode isn't a vector.
17064 (aarch64_output_scalar_simd_mov_immediate): Likewise.
17065 (aarch64_expand_mov_immediate): Update calls after above changes.
17066 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
17067 (aarch64_and_bitmask_imm): Check for scalar integer modes.
17068 (aarch64_move_imm): Likewise.
17069 (aarch64_can_const_movi_rtx_p): Likewise.
17070 (aarch64_strip_extend): Likewise.
17071 (aarch64_extr_rtx_p): Likewise.
17072 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
17073 a CONST_INT when the mode parameter is VOIDmode.
17074 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
17076 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17078 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
17079 * stor-layout.c (bitwise_mode_for_mode): Likewise.
17080 (bitwise_type_for_mode): Update accordingly.
17082 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17084 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
17085 * stor-layout.c (mode_for_size_tree): Likewise.
17086 (mode_for_array): Update accordingly.
17087 (layout_decl): Likewise.
17088 (compute_record_mode): Likewise. Only set the mode once.
17090 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17092 * target.def (get_mask_mode): Change return type to opt_mode.
17094 * doc/tm.texi: Regenerate.
17095 * targhooks.h (default_get_mask_mode): Return an opt_mode.
17096 * targhooks.c (default_get_mask_mode): Likewise.
17097 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
17098 * optabs-query.c (can_vec_mask_load_store_p): Update use of
17099 targetm.get_mask_mode.
17100 * tree.c (build_truth_vector_type): Likewise.
17102 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17104 * machmode.h (mode_for_vector): Return an opt_mode.
17105 * stor-layout.c (mode_for_vector): Likewise.
17106 (mode_for_int_vector): Update accordingly.
17107 (layout_type): Likewise.
17108 * config/i386/i386.c (emit_memmov): Likewise.
17109 (ix86_expand_set_or_movmem): Likewise.
17110 (ix86_expand_vector_init): Likewise.
17111 (ix86_get_mask_mode): Likewise.
17112 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
17114 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
17115 * expmed.c (extract_bit_field_1): Likewise.
17116 * expr.c (expand_expr_real_2): Likewise.
17117 * optabs-query.c (can_vec_perm_p): Likewise.
17118 (can_vec_mask_load_store_p): Likewise.
17119 * optabs.c (expand_vec_perm): Likewise.
17120 * targhooks.c (default_get_mask_mode): Likewise.
17121 * tree-vect-stmts.c (vectorizable_store): Likewise.
17122 (vectorizable_load): Likewise.
17123 (get_vectype_for_scalar_type_and_size): Likewise.
17125 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17127 * machmode.h (mode_for_int_vector): New function.
17128 * stor-layout.c (mode_for_int_vector): Likewise.
17129 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
17130 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
17131 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
17132 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
17133 (s390_expand_vcond): Likewise.
17135 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17137 * machmode.h (opt_machine_mode): New type.
17138 (opt_mode<T>): Allow construction from anything that can be
17140 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
17141 (mode_for_size): Return an opt_machine_mode.
17142 * stor-layout.c (mode_for_size): Likewise.
17143 (mode_for_size_tree): Update call accordingly.
17144 (bitwise_mode_for_mode): Likewise.
17145 (make_fract_type): Likewise.
17146 (make_accum_type): Likewise.
17147 * caller-save.c (replace_reg_with_saved_mem): Update call
17149 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17150 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17151 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17152 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
17153 * expmed.c (extract_bit_field_1): Likewise.
17154 * reload.c (get_secondary_mem): Likewise.
17155 * varasm.c (assemble_integer): Likewise.
17156 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
17159 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17161 * machmode.h (decimal_float_mode_for_size): New function.
17162 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
17163 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
17164 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
17165 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
17166 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
17167 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
17169 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17171 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
17172 (get_builtin_sync_mode): Likewise.
17173 (expand_ifn_atomic_compare_exchange): Likewise.
17174 (expand_builtin_atomic_clear): Likewise.
17175 (expand_builtin_atomic_test_and_set): Likewise.
17176 (fold_builtin_atomic_always_lock_free): Likewise.
17177 * calls.c (compute_argument_addresses): Likewise.
17178 (emit_library_call_value_1): Likewise.
17179 (store_one_arg): Likewise.
17180 * combine.c (combine_instructions): Likewise.
17181 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
17182 * config/arm/arm.c (arm_function_value): Likewise.
17183 (aapcs_allocate_return_reg): Likewise.
17184 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
17185 * config/i386/i386.c (construct_container): Likewise.
17186 (ix86_gimplify_va_arg): Likewise.
17187 (ix86_expand_sse_cmp): Likewise.
17188 (emit_memmov): Likewise.
17189 (emit_memset): Likewise.
17190 (expand_small_movmem_or_setmem): Likewise.
17191 (ix86_expand_pextr): Likewise.
17192 (ix86_expand_pinsr): Likewise.
17193 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
17194 * config/microblaze/microblaze.c (microblaze_block_move_straight):
17196 * config/mips/mips.c (mips_function_value_1) Likewise.
17197 (mips_block_move_straight): Likewise.
17198 (mips_expand_ins_as_unaligned_store): Likewise.
17199 * config/powerpcspe/powerpcspe.c
17200 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17201 (rs6000_darwin64_record_arg_flush): Likewise.
17202 * config/rs6000/rs6000.c
17203 (rs6000_darwin64_record_arg_advance_flush): Likewise.
17204 (rs6000_darwin64_record_arg_flush): Likewise.
17205 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
17206 (sparc_function_value_1): Likewise.
17207 * config/spu/spu.c (adjust_operand): Likewise.
17208 (spu_emit_branch_or_set): Likewise.
17209 (arith_immediate_p): Likewise.
17210 * emit-rtl.c (gen_lowpart_common): Likewise.
17211 * expr.c (expand_expr_real_1): Likewise.
17212 * function.c (assign_parm_setup_block): Likewise.
17213 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
17214 * reload1.c (alter_reg): Likewise.
17215 * stor-layout.c (mode_for_vector): Likewise.
17216 (layout_type): Likewise.
17218 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
17220 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
17221 (spu_convert_move): Likewise.
17222 * lower-subreg.c (resolve_simple_move): Likewise.
17224 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17227 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
17228 define_insn to a define_expand.
17229 (altivec_vsum2sws_direct): New define_insn.
17230 (altivec_vsumsws): Convert from a define_insn to a define_expand.
17232 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17234 * config/arm/arm.c (arm_option_params_internal): Improve setting of
17237 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
17242 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
17243 realign stack if stack alignment needed is less than incoming
17246 2017-09-05 Marek Polacek <polacek@redhat.com>
17249 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
17252 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
17254 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
17256 2017-09-05 Richard Biener <rguenther@suse.de>
17258 PR tree-optimization/82084
17259 * fold-const.c (can_native_encode_string_p): Handle wide characters.
17261 2017-09-05 Richard Biener <rguenther@suse.de>
17263 PR tree-optimization/82102
17264 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
17266 2017-09-05 Martin Liska <mliska@suse.cz>
17268 PR tree-optimization/82032
17269 * tree-cfg.c (generate_range_test): New function.
17270 * tree-cfg.h (generate_range_test): Declared here.
17271 * tree-cfgcleanup.c (convert_single_case_switch): New function.
17272 (cleanup_control_expr_graph): Use it.
17273 * tree-switch-conversion.c (try_switch_expansion): Remove
17275 (emit_case_nodes): Use generate_range_test.
17277 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17280 * config/i386/i386.md (*<btsc><mode>_mask): Add
17281 TARGET_USE_BT to insn constraint.
17282 (*btr<mode>_mask): Ditto.
17284 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
17286 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
17287 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
17289 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
17292 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
17293 TARGET_NEON and TARGET_IWMMXT.
17294 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
17295 TARGET_NEON and TARGET_IWMMXT.
17296 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
17298 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
17300 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
17301 (ix86_rewrite_tls_address): Ditto.
17302 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
17303 (ix86_rewrite_tls_address_1): Ditto.
17304 (ix86_rewrite_tls_address): Ditto.
17305 * config/i386/predicates.md (tls_address_pattern): New predicate.
17306 * config/i386/i386.md (TLS address splitter): New splitter.
17308 2017-09-04 Richard Biener <rguenther@suse.de>
17310 PR tree-optimization/82084
17311 * fold-const.h (can_native_encode_string_p): Declare.
17312 * fold-const.c (can_native_encode_string_p): Factor out from ...
17313 (native_encode_string): ... here.
17314 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
17315 vectorizing stores from constants we later cannot handle.
17317 2017-09-04 Marek Polacek <polacek@redhat.com>
17320 * doc/invoke.texi: Update -Wtautological-compare documentation.
17322 2017-09-04 Jeff Law <law@redhat.com>
17324 PR tree-optimization/64910
17325 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
17326 swap the first and last operand if the last is a constant.
17328 2017-09-04 Marek Polacek <polacek@redhat.com>
17331 * convert.c (do_narrow): When sanitizing signed integer overflows,
17332 bail out for signed types.
17333 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
17335 2017-09-04 Richard Biener <rguenther@suse.de>
17337 PR tree-optimization/82060
17338 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
17339 Move devirtualization after stmt folding and before EH/AB/noreturn
17340 cleanup to get the stmt refs canonicalized. Use a bool instead
17341 of gimple_modified_p since that doesn't work for NOPs. Schedule
17342 NOPs generated by folding for removal.
17344 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17345 Alan Hayward <alan.hayward@arm.com>
17346 David Sherwood <david.sherwood@arm.com>
17348 * coretypes.h (pad_direction): New enum.
17349 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
17350 (FUNCTION_ARG_PADDING): Likewise.
17351 * target.def (function_arg_padding): New hook.
17352 * targhooks.h (default_function_arg_padding): Declare.
17353 * targhooks.c (default_function_arg_padding): New function.
17354 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
17355 (TARGET_FUNCTION_ARG_PADDING): ...this.
17356 * doc/tm.texi: Regenerate.
17357 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
17358 instead of direction.
17359 (compute_argument_addresses): Likewise.
17360 (load_register_parameters): Likewise.
17361 (emit_library_call_value_1): Likewise.
17362 (store_one_arg): Use targetm.calls.function_arg_padding instead
17363 of FUNCTION_ARG_PADDING.
17364 (must_pass_in_stack_var_size_or_pad): Likewise.
17365 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
17366 (emit_group_store): Likewise.
17367 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
17368 instead of FUNCTION_ARG_PADDING.
17369 (emit_push_insn): Likewise, and propagate enum change throughout
17371 * function.h (direction): Delete.
17372 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
17374 * function.c (assign_parm_find_stack_rtl): Likewise.
17375 (assign_parm_setup_block_p): Likewise.
17376 (assign_parm_setup_block): Likewise.
17377 (gimplify_parameters): Likewise.
17378 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
17379 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
17381 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
17382 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17383 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
17384 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
17385 (aarch64_function_arg_padding): ...this new function.
17386 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
17387 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17388 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
17389 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17390 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
17391 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17392 (arm_pad_arg_upward): Replace with...
17393 (arm_function_arg_padding): ...this new function.
17394 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
17396 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
17397 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
17398 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17399 (ia64_hpux_function_arg_padding): Replace with...
17400 (ia64_function_arg_padding): ...this new function. Use pad_direction
17401 instead of direction. Check for TARGET_HPUX.
17402 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
17403 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17404 (iq2000_function_arg_padding): New function.
17405 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
17406 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
17407 (mips_function_arg_padding): ...this new function.
17408 (mips_pad_reg_upward): Update accordingly.
17409 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17410 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
17411 targetm.calls.function_arg_padding.
17412 (FUNCTION_ARG_PADDING): Delete.
17413 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
17414 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
17415 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17416 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
17417 (nios2_block_reg_padding): Return pad_direction instead of direction.
17418 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
17419 instead of direction.
17420 (nios2_function_arg_padding): Likewise. Make static.
17421 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17422 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
17423 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
17424 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
17425 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17426 (pa_function_arg_padding): Make static. Return pad_direction instead
17428 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
17429 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17430 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
17431 instead of direction. Use targetm.calls.function_arg_padding.
17432 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
17433 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17434 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
17435 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
17436 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
17438 (function_arg_padding): Rename to...
17439 (rs6000_function_arg_padding): ...this. Make static. Return
17440 pad_direction instead of direction.
17441 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17442 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
17443 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
17444 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
17445 instead of direction. Use targetm.calls.function_arg_padding.
17446 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
17447 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
17448 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
17449 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
17450 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17451 (function_arg_padding): Rename to...
17452 (rs6000_function_arg_padding): ...this. Make static. Return
17453 pad_direction instead of direction.
17454 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
17455 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
17456 * config/s390/s390.c (s390_function_arg_padding): New function.
17457 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17458 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
17459 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
17460 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
17461 (function_arg_padding): Rename to...
17462 (sparc_function_arg_padding): ...this. Make static. Return
17463 pad_direction instead of direction.
17464 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
17465 * config/spu/spu.c (spu_function_arg_padding): New function.
17466 (TARGET_FUNCTION_ARG_PADDING): Redefine.
17467 * system.h (FUNCTION_ARG_PADDING): Poison.
17469 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17470 Alan Hayward <alan.hayward@arm.com>
17471 David Sherwood <david.sherwood@arm.com>
17473 * target.def (modes_tieable_p): New hook.
17474 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
17475 (TARGET_MODES_TIEABLE_P): ...this.
17476 * doc/tm.texi.in: Regenerate.
17477 * hooks.h (hook_bool_mode_mode_true): Declare.
17478 * hooks.c (hook_bool_mode_mode_true): New function.
17479 * combine.c (subst): Use targetm.modes_tieable_p instead of
17481 * dse.c (find_shift_sequence): Likewise.
17482 * expmed.c (extract_low_bits): Likewise.
17483 * lower-subreg.c: Include target.h.
17484 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
17486 * rtlanal.c (rtx_cost): Likewise.
17487 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
17488 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
17489 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
17490 (TARGET_MODES_TIEABLE_P): Redefine.
17491 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
17492 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
17493 (TARGET_MODES_TIEABLE_P): Redefine.
17494 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
17495 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
17496 (arc_modes_tieable_p): New function.
17497 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
17498 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
17499 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
17500 (arm_modes_tieable_p): Make static.
17501 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
17502 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
17503 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
17504 (TARGET_MODES_TIEABLE_P): Redefine.
17505 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
17506 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
17507 (TARGET_MODES_TIEABLE_P): Redefine.
17508 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
17509 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
17510 (cr16_modes_tieable_p): New function.
17511 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
17512 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
17513 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
17514 (TRULY_NOOP_TRUNCATION): Update comment.
17515 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
17516 (TRULY_NOOP_TRUNCATION): Update comment.
17517 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
17518 (frv_modes_tieable_p): New function.
17519 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
17520 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
17521 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
17522 (TARGET_MODES_TIEABLE_P): Redefine.
17523 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
17524 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
17525 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
17526 (TARGET_MODES_TIEABLE_P): Redefine.
17527 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
17528 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
17529 (ia64_modes_tieable_p): New function.
17530 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
17531 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
17532 (iq2000_modes_tieable_p): New function.
17533 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
17534 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
17535 (lm32_modes_tieable_p): New function.
17536 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
17537 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
17538 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
17539 (TARGET_MODES_TIEABLE_P): Redefine.
17540 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
17541 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
17542 (m32r_modes_tieable_p): New function.
17543 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
17544 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
17545 (m68k_modes_tieable_p): New function.
17546 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
17547 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
17548 (mcore_modes_tieable_p): New function.
17549 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
17550 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
17552 (TARGET_MODES_TIEABLE_P): Redefine.
17553 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
17554 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
17555 * config/mips/mips.c (mips_modes_tieable_p): Make static.
17556 (TARGET_MODES_TIEABLE_P): Redefine.
17557 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
17558 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
17559 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
17560 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
17561 (mn10300_modes_tieable_p): ...this and make static.
17562 (TARGET_MODES_TIEABLE_P): Redefine.
17563 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
17564 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
17565 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
17566 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
17567 (msp430_modes_tieable_p): Make static.
17568 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
17569 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
17570 (TARGET_MODES_TIEABLE_P): Redefine.
17571 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
17572 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
17573 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
17574 (TARGET_MODES_TIEABLE_P): Redefine.
17575 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
17576 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
17577 * config/pa/pa.c (pa_modes_tieable_p): Make static.
17578 (TARGET_MODES_TIEABLE_P): Redefine.
17579 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
17580 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
17581 (pdp11_modes_tieable_p): New function.
17582 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
17583 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
17584 (rs6000_modes_tieable_p): New function.
17585 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17586 * config/powerpcspe/powerpcspe.md: Update comment.
17587 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
17588 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
17589 (TARGET_MODES_TIEABLE_P): Redefine.
17590 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
17591 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
17592 (rl78_modes_tieable_p): New function.
17593 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
17594 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
17595 (rs6000_modes_tieable_p): New function.
17596 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
17597 * config/rs6000/rs6000.md: Update comment.
17598 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
17599 * config/rx/rx.c (rx_modes_tieable_p): New function.
17600 (TARGET_MODES_TIEABLE_P): Redefine.
17601 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
17602 * config/s390/s390.c (s390_modes_tieable_p): New function.
17603 (TARGET_MODES_TIEABLE_P): Redefine.
17604 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
17605 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
17606 (sh_modes_tieable_p): New function.
17607 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
17608 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
17609 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
17610 (sparc_modes_tieable_p): Make static.
17611 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
17612 * config/spu/spu.c (spu_modes_tieable_p): New function.
17613 (TARGET_MODES_TIEABLE_P): Redefine.
17614 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
17615 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
17616 (TARGET_MODES_TIEABLE_P): Redefine.
17617 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
17618 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
17619 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
17620 * config/v850/v850.c (v850_modes_tieable_p): New function.
17621 (TARGET_MODES_TIEABLE_P): Redefine.
17622 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
17623 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
17624 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
17625 (visium_modes_tieable_p): New function.
17626 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
17627 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
17628 (xtensa_modes_tieable_p): New function.
17629 * system.h (MODES_TIEABLE_P): Poison.
17631 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17632 Alan Hayward <alan.hayward@arm.com>
17633 David Sherwood <david.sherwood@arm.com>
17635 * target.def (hard_regno_mode_ok): New hook.
17636 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
17637 (TARGET_HARD_REGNO_MODE_OK): ...this.
17638 * doc/tm.texi.in: Regenerate.
17639 * hooks.h (hook_bool_uint_mode_true): Declare.
17640 * hooks.c (hook_bool_uint_mode_true): New function.
17641 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
17642 HARD_REGNO_MODE_OK.
17643 * genpreds.c (write_insn_preds_c): Add an include of target.h.
17644 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
17645 instead of HARD_REGNO_MODE_OK.
17646 * caller-save.c: Include target.h.
17647 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
17648 HARD_REGNO_MODE_OK.
17649 * combine.c (can_combine_p): Likewise.
17650 (combinable_i3pat): Likewise.
17651 (can_change_dest_mode): Likewise.
17652 * expr.c (init_expr_target): Likewise.
17653 (convert_move): Likewise.
17654 (convert_modes): Likewise.
17655 * ira.c (setup_prohibited_class_mode_regs): Likewise.
17656 (setup_prohibited_mode_move_regs): Likewise.
17657 * ira.h (target_ira): Likewise.
17658 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17659 * lra-constraints.c (process_alt_operands): Likewise.
17660 (split_reg): Likewise.
17661 * recog.c (peep2_find_free_register): Likewise.
17662 * ree.c (combine_reaching_defs): Likewise.
17663 * regcprop.c (maybe_mode_change): Likewise.
17664 * reginfo.c (init_reg_sets_1): Likewise.
17665 (choose_hard_reg_mode): Likewise.
17666 (simplifiable_subregs): Likewise.
17667 * regrename.c (check_new_reg_p): Likewise.
17668 * reload.c (find_valid_class): Likewise.
17669 (find_valid_class_1): Likewise.
17670 (reload_inner_reg_of_subreg): Likewise.
17671 (push_reload): Likewise.
17672 (combine_reloads): Likewise.
17673 (find_dummy_reload): Likewise.
17674 (find_reloads): Likewise.
17675 * reload1.c (find_reg): Likewise.
17676 (set_reload_reg): Likewise.
17677 (allocate_reload_reg): Likewise.
17678 (choose_reload_regs): Likewise.
17679 (reload_adjust_reg_for_temp): Likewise.
17680 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
17681 (simplify_subreg_regno): Likewise.
17682 * sel-sched.c (init_regs_for_mode): Likewise.
17683 * varasm.c (make_decl_rtl): Likewise.
17684 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
17685 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
17686 HARD_REGNO_MODE_OK.
17687 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
17688 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
17689 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17690 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
17691 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
17692 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17693 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
17694 (arc_mode_class): Delete.
17695 (HARD_REGNO_MODE_OK): Delete.
17696 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17697 (arc_hard_regno_mode_ok): Rename old array to...
17698 (arc_hard_regno_mode_ok_modes): ...this.
17699 (arc_conditional_register_usage): Update accordingly.
17700 (arc_mode_class): Make static.
17701 (arc_hard_regno_mode_ok): New function.
17702 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
17703 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
17704 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17705 (arm_hard_regno_mode_ok): Make static.
17706 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
17707 HARD_REGNO_MODE_OK.
17708 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
17709 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
17710 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
17712 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17713 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
17714 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
17715 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
17716 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
17717 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17718 * config/bfin/predicates.md (valid_reg_operand): Use
17719 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
17720 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
17721 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
17722 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17723 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
17724 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
17725 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17726 (cr16_hard_regno_mode_ok): Make static and return a bool.
17727 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
17728 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17729 (cris_hard_regno_mode_ok): New function.
17730 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
17731 (epiphany_mode_class): Delete.
17732 (HARD_REGNO_MODE_OK): Delete.
17733 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
17734 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17735 (hard_regno_mode_ok): Rename to...
17736 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
17737 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
17738 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
17739 HARD_REGNO_MODE_OK.
17740 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
17741 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
17742 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17743 (frv_hard_regno_mode_ok): Make static and return a bool.
17744 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
17745 HARD_REGNO_MODE_OK.
17746 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
17747 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
17748 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
17749 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
17751 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17752 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
17753 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
17754 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
17756 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17757 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
17758 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17759 (ia64_hard_regno_mode_ok): New function.
17760 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
17761 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17762 (iq2000_hard_regno_mode_ok): New function.
17763 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
17764 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17765 (lm32_hard_regno_mode_ok): New function.
17766 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
17767 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
17768 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
17769 instead of HARD_REGNO_MODE_OK.
17770 (m32c_hard_regno_ok): Rename to...
17771 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
17772 (m32c_cannot_change_mode_class): Update accordingly.
17773 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17774 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
17775 (m32r_mode_class): Delete.
17776 (HARD_REGNO_MODE_OK): Delete.
17777 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17778 (m32r_hard_regno_mode_ok): Rename to...
17779 (m32r_hard_regno_modes): ...this.
17780 (m32r_mode_class): Make static.
17781 (m32r_hard_regno_mode_ok): New function.
17782 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
17783 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
17784 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17785 (m68k_hard_regno_mode_ok): Make static.
17786 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
17787 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17788 (mcore_hard_regno_mode_ok): New function.
17789 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
17790 (HARD_REGNO_MODE_OK): Delete.
17791 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
17793 (microblaze_hard_regno_mode_ok_p): ...this and make static.
17794 (microblaze_hard_regno_mode_ok): New function.
17795 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17796 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
17797 (mips_hard_regno_mode_ok): Delete.
17798 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
17799 (mips_hard_regno_mode_ok_p): ...this and make static.
17800 (mips_hard_regno_mode_ok_p): Rename to...
17801 (mips_hard_regno_mode_ok_uncached): ...this.
17802 (mips_hard_regno_mode_ok): New function.
17803 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
17804 of HARD_REGNO_MODE_OK.
17805 (mips_option_override): Update after above name changes.
17806 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17807 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
17808 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
17809 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
17810 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
17811 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17812 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
17813 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
17814 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
17815 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17816 (msp430_hard_regno_mode_ok): Make static and return a bool.
17817 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
17818 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
17819 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
17821 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17822 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
17823 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
17824 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
17825 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
17826 (PA_HARD_REGNO_MODE_OK): ...this
17827 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
17828 (PA_HARD_REGNO_MODE_OK): ...this.
17829 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17830 (pa_hard_regno_mode_ok): New function.
17831 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
17832 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17833 (pdp11_hard_regno_mode_ok): New function.
17834 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
17835 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
17837 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
17839 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17840 (rs6000_hard_regno_mode_ok): Rename to...
17841 (rs6000_hard_regno_mode_ok_uncached): ...this.
17842 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17843 (rs6000_hard_regno_mode_ok): New function.
17844 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
17845 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
17846 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
17847 (riscv_hard_regno_mode_ok): ...this and make static.
17848 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17849 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
17850 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
17851 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17852 (rl78_hard_regno_mode_ok): Make static and return bool.
17853 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
17854 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
17856 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
17857 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17858 (rs6000_hard_regno_mode_ok): Rename to...
17859 (rs6000_hard_regno_mode_ok_uncached): ...this.
17860 (rs6000_init_hard_regno_mode_ok): Update accordingly.
17861 (rs6000_hard_regno_mode_ok): New function.
17862 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
17863 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
17864 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17865 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
17866 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
17867 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
17868 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17869 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
17870 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
17871 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17872 (sh_hard_regno_mode_ok): Make static.
17873 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
17874 instead of HARD_REGNO_MODE_OK.
17875 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
17876 (sparc_mode_class): Delete.
17877 (HARD_REGNO_MODE_OK): Delete.
17878 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17879 (hard_regno_mode_classes): Make static.
17880 (sparc_mode_class): Likewise.
17881 (sparc_hard_regno_mode_ok): New function.
17882 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
17883 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
17884 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
17886 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17887 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
17888 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
17889 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
17890 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
17891 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17892 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
17893 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
17894 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
17895 (visium_hard_regno_mode_ok): New function.
17896 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
17897 instead of HARD_REGNO_MODE_OK.
17898 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
17899 (HARD_REGNO_MODE_OK): Delete.
17900 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
17901 (xtensa_hard_regno_mode_ok_p): ...this and make static.
17902 (xtensa_option_override): Update accordingly.
17903 (TARGET_HARD_REGNO_MODE_OK): Redefine.
17904 (xtensa_hard_regno_mode_ok): New function.
17905 * system.h (HARD_REGNO_MODE_OK): Poison.
17907 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17908 Alan Hayward <alan.hayward@arm.com>
17909 David Sherwood <david.sherwood@arm.com>
17911 * target.def (hard_regno_call_part_clobbered): New hook.
17912 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
17913 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
17914 * doc/tm.texi: Regenerate.
17915 * hooks.h (hook_bool_uint_mode_false): Declare.
17916 * hooks.c (hook_bool_uint_mode_false): New function.
17917 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17918 * cselib.c (cselib_process_insn): Use
17919 targetm.hard_regno_call_part_clobbered instead of
17920 HARD_REGNO_CALL_PART_CLOBBERED.
17921 * ira-conflicts.c (ira_build_conflicts): Likewise.
17922 * ira-costs.c (ira_tune_allocno_costs): Likewise.
17923 * lra-constraints.c (need_for_call_save_p): Likewise.
17924 * lra-lives.c: Include target.h.
17925 (check_pseudos_live_through_calls): Use
17926 targetm.hard_regno_call_part_clobbered instead of
17927 HARD_REGNO_CALL_PART_CLOBBERED.
17928 * regcprop.c: Include target.h.
17929 (copyprop_hardreg_forward_1): Use
17930 targetm.hard_regno_call_part_clobbered instead of
17931 HARD_REGNO_CALL_PART_CLOBBERED.
17932 * reginfo.c (choose_hard_reg_mode): Likewise.
17933 * regrename.c (check_new_reg_p): Likewise.
17934 * reload.c (find_equiv_reg): Likewise.
17935 * reload1.c (emit_reload_insns): Likewise.
17936 * sched-deps.c (deps_analyze_insn): Likewise.
17937 * sel-sched.c (init_regs_for_mode): Likewise.
17938 (mark_unavailable_hard_regs): Likewise.
17939 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
17940 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17941 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
17943 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17944 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17945 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
17947 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
17949 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17950 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17951 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
17953 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17954 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17955 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
17957 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17958 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
17960 * config/powerpcspe/powerpcspe.c
17961 (rs6000_hard_regno_call_part_clobbered): New function.
17962 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17963 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17964 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
17966 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17967 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17968 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
17970 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
17971 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
17972 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
17974 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
17975 Alan Hayward <alan.hayward@arm.com>
17976 David Sherwood <david.sherwood@arm.com>
17978 * rtl.h (subreg_memory_offset): Declare.
17979 * emit-rtl.c (subreg_memory_offset): New function.
17980 * expmed.c (store_bit_field_1): Use it.
17981 * expr.c (undefined_operand_subword_p): Likewise.
17982 * simplify-rtx.c (simplify_subreg): Likewise.
17984 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
17986 PR rtl-optimization/57448
17989 * optabs.c (expand_atomic_load): Place compiler memory barriers if
17990 using atomic_load pattern.
17991 (expand_atomic_store): Likewise.
17993 2017-09-04 Jakub Jelinek <jakub@redhat.com>
17996 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
17997 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
17998 handling. Use replace_call_with_value with NULL instead of
17999 gsi_replace, unlink_stmt_vdef and release_defs.
18001 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
18004 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
18006 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
18009 * rtl.h (emit_library_call_value_1): Declare.
18010 (emit_library_call): Replace declaration with a series of overloads.
18011 Remove the parameter count argument.
18012 (emit_library_call_value): Likewise.
18013 * calls.c (emit_library_call_value_1): Make global. Replace varargs
18014 with an "rtx_mode_t *".
18015 (emit_library_call_value): Delete.
18016 (emit_library_call): Likewise.
18017 * asan.c (asan_emit_stack_protection): Update calls accordingly.
18018 (asan_emit_allocas_unpoison): Likewise.
18019 * builtins.c (expand_builtin_powi): Likewise.
18020 (expand_asan_emit_allocas_unpoison): Likewise.
18021 * cfgexpand.c (expand_main_function): Likewise.
18022 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
18023 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
18024 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
18025 * config/arm/arm.c (arm_trampoline_init): Likewise.
18026 (arm_call_tls_get_addr): Likewise.
18027 (arm_expand_divmod_libfunc): Likewise.
18028 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
18029 (smulsi3_highpart): Likewise.
18030 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
18031 (c6x_expand_compare): Likewise.
18032 (c6x_expand_movmem): Likewise.
18033 * config/frv/frv.c (frv_trampoline_init): Likewise.
18034 * config/i386/i386.c (ix86_trampoline_init): Likewise.
18035 (ix86_expand_divmod_libfunc): Likewise.
18036 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
18037 (ia64_expand_compare): Likewise.
18038 (ia64_profile_hook): Likewise.
18039 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
18040 (nonlocal_goto): Likewise.
18041 (restore_stack_nonlocal): Likewise.
18042 * config/m32r/m32r.c (block_move_call): Likewise.
18043 (m32r_trampoline_init): Likewise.
18044 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
18045 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
18046 (m68k_call_m68k_read_tp): Likewise.
18047 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
18048 (microblaze_expand_divide): Likewise.
18049 * config/mips/mips.h (mips_args): Likewise.
18050 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
18051 (MIPS_ICACHE_SYNC): Likewise.
18052 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
18053 (nios2_trampoline_init): Likewise.
18054 * config/pa/pa.c (hppa_tls_call): Likewise.
18055 (pa_trampoline_init): Likewise.
18056 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
18057 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
18058 (expand_strn_compare): Likewise.
18059 (rs6000_generate_compare): Likewise.
18060 (rs6000_expand_float128_convert): Likewise.
18061 (output_profile_hook): Likewise.
18062 (rs6000_trampoline_init): Likewise.
18063 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
18064 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
18065 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
18066 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
18067 (rs6000_generate_compare): Likewise.
18068 (rs6000_expand_float128_convert): Likewise.
18069 (output_profile_hook): Likewise.
18070 (rs6000_trampoline_init): Likewise.
18071 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
18072 * config/sh/sh.c (sh_trampoline_init): Likewise.
18073 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
18074 (sparc_emit_float_lib_cmp): Likewise.
18075 (sparc32_initialize_trampoline): Likewise.
18076 (sparc64_initialize_trampoline): Likewise.
18077 (sparc_profile_hook): Likewise.
18078 * config/spu/spu.c (ea_load_store): Likewise.
18079 * config/spu/spu.md (floatunssidf2): Likewise.
18080 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
18081 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
18082 * config/visium/visium.c (expand_block_move_4): Likewise.
18083 (expand_block_move_2): Likewise.
18084 (expand_block_move_1): Likewise.
18085 (expand_block_set_4): Likewise.
18086 (expand_block_set_2): Likewise.
18087 (expand_block_set_1): Likewise.
18088 (visium_trampoline_init): Likewise.
18089 (visium_profile_hook): Likewise.
18090 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
18091 (xtensa_setup_frame_addresses): Likewise.
18092 (xtensa_trampoline_init): Likewise.
18093 * except.c (sjlj_emit_function_enter): Likewise.
18094 (sjlj_emit_function_exit): Likewise.
18095 * explow.c (allocate_dynamic_stack_space): Likewise.
18096 (probe_stack_range): Likewise.
18097 * expr.c (convert_mode_scalar): Likewise.
18098 * optabs.c (expand_binop): Likewise.
18099 (expand_twoval_binop_libfunc): Likewise.
18100 (expand_unop): Likewise.
18101 (prepare_cmp_insn): Likewise.
18102 (prepare_float_lib_cmp): Likewise.
18103 (expand_float): Likewise.
18104 (expand_fix): Likewise.
18105 (expand_fixed_convert): Likewise.
18106 (maybe_emit_sync_lock_test_and_set): Likewise.
18107 (expand_atomic_compare_and_swap): Likewise.
18108 (expand_mem_thread_fence): Likewise.
18109 (expand_atomic_fetch_op): Likewise.
18111 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
18113 * doc/generic.texi (OpenACC): Adjust URL.
18114 * doc/invoke.texi (C Dialect Options): Ditto.
18116 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
18118 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
18119 predicate for operand 1. Add (m,<S>) constraint.
18120 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
18121 Prevent memory operand 1 with register operand 2.
18123 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
18125 PR rtl-optimization/82024
18126 * combine.c (try_combine): If the combination result is a PARALLEL,
18127 and we only need to retain the SET in there that would be placed
18128 at I2, check that we can place that at I3 instead, before doing so.
18130 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18133 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
18135 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
18136 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
18139 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
18140 Jeff Law <law@redhat.com>
18142 * varasm.c (bss_initializer_p): Do not put constants into .bss
18143 (categorize_decl_for_section): Handle bss_initializer_p returning
18144 false when DECL_INITIAL is NULL.
18146 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18149 * config/s390/s390.c (s390_can_inline_p): New function.
18151 2017-09-01 Jeff Law <law@redhat.com>
18153 PR tree-optimization/82052
18154 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
18155 Always initialize the returned slot after a hash table miss
18156 when INSERT is true.
18158 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
18160 * config/s390/s390.md (mem_signal_fence): Remove.
18161 * doc/md.texi (mem_signal_fence): Remove.
18162 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
18164 * target-insns.def (mem_signal_fence): Remove.
18166 2017-09-01 Jakub Jelinek <jakub@redhat.com>
18169 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
18172 * asan.c (create_odr_indicator): Strip name encoding from assembler
18173 name before appending it after __odr_asan_.
18175 2017-09-01 Martin Liska <mliska@suse.cz>
18177 PR tree-optimization/82059
18178 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
18179 frequency only when an edge is redirected.
18181 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18183 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
18184 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
18185 (arc_conditional_register_usage): Remove ARC600 lp_count
18187 (arc_file_start): Emit Tag_ARC_CPU_variation.
18188 (arc_can_use_doloop_p): New conditions to use ZOLs.
18189 (hwloop_fail): New function.
18190 (hwloop_optimize): Likewise.
18191 (hwloop_pattern_reg): Likewise.
18192 (arc_doloop_hooks): New struct, to be used with reorg_loops.
18193 (arc_reorg_loops): New function, calls reorg_loops.
18194 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
18195 (arc600_corereg_hazard): Remove ZOL checking, case handled by
18197 (arc_loop_hazard): Remove function, functionality moved into
18199 (arc_hazard): Remove arc_loop_hazard call.
18200 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
18201 into hwloop_optimize.
18202 (arc_label_align): Remove ZOL handling.
18203 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
18204 * config/arc/arc.md (doloop_begin): Remove pattern.
18205 (doloop_begin_i): Likewise.
18206 (doloop_end_i): Likewise.
18207 (doloop_fallback): Likewise.
18208 (doloop_fallback_m): Likewise.
18209 (doloop_end): Reimplement expand.
18210 (arc_lp): New pattern for LP instruction.
18211 (loop_end): New pattern.
18212 (loop_fail): Likewise.
18213 (decrement_and_branch_until_zero): Likewise.
18214 * config/arc/arc.opt (mlpc-width): New option.
18215 * doc/invoke.texi (mlpc-width): Document option.
18217 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18219 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
18220 (arc_ccfsm_advance): Fix checking for delay slots.
18221 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
18223 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18225 * config/arc/arc.md (movqi_insn): Add stores to save constant long
18227 (movhi_insn): Update store instruction constraint which are saving
18228 6-bit short immediates.
18229 (movsi_insn): Consider also short scaled load operations.
18230 (zero_extendhisi2_i): Use Usd constraint instead of T.
18231 (extendhisi2_i): Add q constraint.
18232 (arc_clzsi2): Add type and length attributes.
18233 (arc_ctzsi2): Likewise.
18234 * config/arc/constraints.md (Usc): Update constraint, the
18235 assembler can parse two relocations for a single instruction.
18237 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
18239 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
18240 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
18242 2017-08-31 Olivier Hainque <hainque@adacore.com>
18244 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
18245 match as powerpc-wrs-vxworks*.
18247 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
18249 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
18250 register constraint for by-element operand.
18251 (aarch64_mls_elt_merge<mode>): Likewise.
18253 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18255 * config/arc/arc.c (arc_can_follow_jump): Check for short
18258 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18260 * config.gcc: Use g.opt for arc.
18261 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
18262 functionality moved to ...
18263 (legitimate_scaled_address_p): New function, ...here.
18264 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
18265 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
18266 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
18268 (arc_override_options): Handle G option.
18269 (arc_output_pic_addr_const): Correct function definition.
18270 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
18271 (arc_decl_anon_ns_mem_p): Delete.
18272 (arc_in_small_data_p): Overhaul this function to take into
18273 consideration the value given via G option.
18274 (arc_rewrite_small_data_1): Renamed and corrected old
18275 arc_rewrite_small_data function.
18276 (arc_rewrite_small_data): New function.
18277 (small_data_pattern): Don't use pic_offset_table_rtx.
18278 * config/arc/arc.h (CC1_SPEC): Recognize G option.
18279 * config/arc/simdext.md (movmisalignv2hi): Use
18280 prepare_move_operands function.
18282 (movmisalign*): Likewise.
18283 * doc/invoke.texi (ARC options): Document -G option.
18285 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
18287 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
18289 * config/arc/arc.c (arc_print_operand): Output scalled address for
18290 sdata whenever is possible.
18291 (arc_in_small_data_p): Allow sdata for 64bit datum when double
18292 load/stores are available.
18293 (compact_sda_memory_operand): Check for the alignment required by
18294 code density instructions.
18295 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
18297 * config/arc/constraints.md (Usd): Update constraint.
18298 (Us0): New constraint.
18299 (Usc): Update constraint.
18301 2017-08-31 Richard Biener <rguenther@suse.de>
18303 PR middle-end/82054
18304 * dwarf2out.c (dwarf2out_early_global_decl): Process each
18305 function only once.
18307 2017-08-31 Tamar Christina <tamar.christina@arm.com>
18309 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
18310 Resize type_signature.
18312 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18313 Alan Hayward <alan.hayward@arm.com>
18314 David Sherwood <david.sherwood@arm.com>
18316 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
18317 subregs whose inner modes can be stored in GPRs.
18318 (aarch64_classify_index): Likewise.
18320 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18321 Alan Hayward <alan.hayward@arm.com>
18322 David Sherwood <david.sherwood@arm.com>
18324 * config/aarch64/iterators.md (V_cmp_result): Rename to...
18325 (V_INT_EQUIV): ...this.
18326 (v_cmp_result): Rename to...
18327 (v_int_equiv): ...this.
18328 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
18329 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
18330 (copysign<mode>3): Likewise.
18331 (aarch64_simd_bsl<mode>_internal): Likewise.
18332 (aarch64_simd_bsl<mode>): Likewise.
18333 (vec_cmp<mode><mode>): Likewise.
18334 (vcond<mode><mode>): Likewise.
18335 (vcond<v_cmp_mixed><mode>): Likewise.
18336 (vcondu<mode><v_cmp_mixed>): Likewise.
18337 (aarch64_cm<optab><mode>): Likewise.
18338 (aarch64_cmtst<mode>): Likewise.
18339 (aarch64_fac<optab><mode>): Likewise.
18340 (vec_perm_const<mode>): Likewise.
18341 (vcond_mask_<mode><v_cmp_result>): Rename to...
18342 (vcond_mask_<mode><v_int_equiv>): ...this.
18343 (vec_cmp<mode><v_cmp_result>): Rename to...
18344 (vec_cmp<mode><v_int_equiv>): ...this.
18346 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
18347 Alan Hayward <alan.hayward@arm.com>
18348 David Sherwood <david.sherwood@arm.com>
18350 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
18352 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
18353 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
18354 (UNSPEC_LD4_DREG): New unspecs.
18355 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
18356 (aarch64_ld2<mode>_dreg_be): Replace with...
18357 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
18359 (aarch64_ld3<mode>_dreg_le)
18360 (aarch64_ld3<mode>_dreg_be): Replace with...
18361 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
18363 (aarch64_ld4<mode>_dreg_le)
18364 (aarch64_ld4<mode>_dreg_be): Replace with...
18365 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
18368 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18370 PR tree-optimization/81987
18371 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
18372 insert an initializer in a location not dominated by the stride
18375 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18377 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
18378 on the entire header of the finally block in the fallthru case.
18380 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
18382 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
18384 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
18386 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
18387 rs6000_emit_move_from_cr and call renamed function.
18388 (rs6000_emit_prologue): Call renamed functions.
18389 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
18390 movesi_from_cr, remove volatile CRs.
18392 2017-08-30 Jon Beniston <jon@beniston.com>
18393 Richard Biener <rguenther@suse.de>
18395 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
18396 of VECTOR_MODE_P check.
18397 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
18398 element vector types.
18400 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18402 * df.h (df_read_modify_subreg_p): Remove in favor of...
18403 * rtl.h (read_modify_subreg_p): ...this new function. Take a
18404 const_rtx instead of an rtx.
18405 * cprop.c (local_cprop_find_used_regs): Update accordingly.
18406 * df-problems.c (df_word_lr_mark_ref): Likewise.
18407 * ira-lives.c (mark_pseudo_reg_live): Likewise.
18408 (mark_pseudo_reg_dead): Likewise.
18409 (mark_ref_dead): Likewise.
18410 * reginfo.c (init_subregs_of_mode): Likewise.
18411 * sched-deps.c (sched_analyze_1): Likewise.
18412 * df-scan.c (df_def_record_1): Likewise.
18413 (df_uses_record): Likewise.
18414 (df_read_modify_subreg_p): Remove in favor of...
18415 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
18416 const_rtx instead of an rtx.
18418 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18419 Alan Hayward <alan.hayward@arm.com>
18420 David Sherwood <david.sherwood@arm.com>
18422 * rtl.h (partial_subreg_p): New function.
18423 * caller-save.c (save_call_clobbered_regs): Use it.
18424 * calls.c (expand_call): Likewise.
18425 * combine.c (combinable_i3pat): Likewise.
18426 (simplify_set): Likewise.
18427 (make_extraction): Likewise.
18428 (make_compound_operation_int): Likewise.
18429 (gen_lowpart_or_truncate): Likewise.
18430 (force_to_mode): Likewise.
18431 (make_field_assignment): Likewise.
18432 (reg_truncated_to_mode): Likewise.
18433 (record_truncated_value): Likewise.
18434 (move_deaths): Likewise.
18435 * cse.c (record_jump_cond): Likewise.
18436 (cse_insn): Likewise.
18437 * cselib.c (cselib_lookup_1): Likewise.
18438 * expmed.c (extract_bit_field_using_extv): Likewise.
18439 * function.c (assign_parm_setup_reg): Likewise.
18440 * ifcvt.c (noce_convert_multiple_sets): Likewise.
18441 * ira-build.c (create_insn_allocnos): Likewise.
18442 * lra-coalesce.c (merge_pseudos): Likewise.
18443 * lra-constraints.c (match_reload): Likewise.
18444 (simplify_operand_subreg): Likewise.
18445 (curr_insn_transform): Likewise.
18446 * lra-lives.c (process_bb_lives): Likewise.
18447 * lra.c (new_insn_reg): Likewise.
18448 (lra_substitute_pseudo): Likewise.
18449 * regcprop.c (mode_change_ok): Likewise.
18450 (maybe_mode_change): Likewise.
18451 (copyprop_hardreg_forward_1): Likewise.
18452 * reload.c (push_reload): Likewise.
18453 (find_reloads): Likewise.
18454 (find_reloads_subreg_address): Likewise.
18455 * reload1.c (alter_reg): Likewise.
18456 (eliminate_regs_1): Likewise.
18457 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
18459 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
18461 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
18462 back to if statements, including unpack.
18464 2017-08-30 Martin Liska <mliska@suse.cz>
18466 PR inline-asm/82001
18467 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
18469 (func_checker::compare_asm_inputs_outputs): ... this function.
18470 (func_checker::compare_gimple_asm): Use the function to compare
18471 also ASM constrains.
18472 * ipa-icf-gimple.h: Rename the function.
18474 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18475 Alan Hayward <alan.hayward@arm.com>
18476 David Sherwood <david.sherwood@arm.com>
18478 * coretypes.h (complex_mode): New type.
18479 * gdbhooks.py (build_pretty_printer): Handle it.
18480 * machmode.h (complex_mode): New class.
18481 (complex_mode::includes_p): New function.
18482 (is_complex_int_mode): Likewise.
18483 (is_complex_float_mode): Likewise.
18484 * genmodes.c (get_mode_class): Handle complex mode classes.
18485 * function.c (expand_function_end): Use is_complex_int_mode.
18487 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18488 Alan Hayward <alan.hayward@arm.com>
18489 David Sherwood <david.sherwood@arm.com>
18491 * coretypes.h (scalar_mode_pod): New typedef.
18492 * gdbhooks.py (build_pretty_printer): Handle it.
18493 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
18494 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
18495 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
18496 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
18497 as_a <scalar_mode>.
18499 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18500 Alan Hayward <alan.hayward@arm.com>
18501 David Sherwood <david.sherwood@arm.com>
18503 * machmode.h (mode_for_vector): Take a scalar_mode instead
18505 * stor-layout.c (mode_for_vector): Likewise.
18506 * explow.c (promote_mode): Use as_a <scalar_mode>.
18507 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
18509 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18510 Alan Hayward <alan.hayward@arm.com>
18511 David Sherwood <david.sherwood@arm.com>
18513 * target.def (preferred_simd_mode): Take a scalar_mode
18514 instead of a machine_mode.
18515 * targhooks.h (default_preferred_simd_mode): Likewise.
18516 * targhooks.c (default_preferred_simd_mode): Likewise.
18517 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
18518 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
18519 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
18520 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
18521 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
18522 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
18523 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
18524 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
18526 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
18527 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
18528 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
18529 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
18530 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
18531 * doc/tm.texi: Regenerate.
18532 * optabs-query.c (can_vec_mask_load_store_p): Return false for
18535 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18536 Alan Hayward <alan.hayward@arm.com>
18537 David Sherwood <david.sherwood@arm.com>
18539 * target.def (scalar_mode_supported_p): Take a scalar_mode
18540 instead of a machine_mode.
18541 * targhooks.h (default_scalar_mode_supported_p): Likewise.
18542 * targhooks.c (default_scalar_mode_supported_p): Likewise.
18543 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
18544 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
18545 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
18546 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
18547 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
18548 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
18549 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
18550 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
18551 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
18552 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
18553 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
18554 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
18556 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
18557 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
18558 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
18559 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
18560 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
18562 * doc/tm.texi: Regenerate.
18564 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18565 Alan Hayward <alan.hayward@arm.com>
18566 David Sherwood <david.sherwood@arm.com>
18568 * coretypes.h (opt_scalar_mode): New typedef.
18569 * gdbhooks.py (build_pretty_printers): Handle it.
18570 * machmode.h (mode_iterator::get_2xwider): Add overload for
18572 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
18574 * expr.c (convert_mode_scalar): Likewise.
18575 * omp-low.c (omp_clause_aligned_alignment): Likewise.
18576 * optabs.c (expand_float): Likewise.
18577 (expand_fix): Likewise.
18578 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18580 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18581 Alan Hayward <alan.hayward@arm.com>
18582 David Sherwood <david.sherwood@arm.com>
18584 * optabs.c (expand_float): Explicitly check for scalars before
18585 using a branching expansion.
18586 (expand_fix): Likewise.
18588 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18589 Alan Hayward <alan.hayward@arm.com>
18590 David Sherwood <david.sherwood@arm.com>
18592 * expr.c (convert_mode): Split scalar handling out into...
18593 (convert_mode_scalar): ...this new function. Treat the modes
18596 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18597 Alan Hayward <alan.hayward@arm.com>
18598 David Sherwood <david.sherwood@arm.com>
18600 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
18602 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
18604 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18605 Alan Hayward <alan.hayward@arm.com>
18606 David Sherwood <david.sherwood@arm.com>
18608 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
18609 rather than a machine_mode.
18610 (fixed_from_string): Likewise.
18611 (fixed_convert): Likewise.
18612 (fixed_convert_from_int): Likewise.
18613 (fixed_convert_from_real): Likewise.
18614 (real_convert_from_fixed): Likewise.
18615 * fixed-value.c (fixed_from_double_int): Likewise.
18616 (fixed_from_string): Likewise.
18617 (fixed_convert): Likewise.
18618 (fixed_convert_from_int): Likewise.
18619 (fixed_convert_from_real): Likewise.
18620 (real_convert_from_fixed): Likewise.
18621 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
18623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18624 Alan Hayward <alan.hayward@arm.com>
18625 David Sherwood <david.sherwood@arm.com>
18627 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
18628 of separate mode class checks. Do not allow vector modes here.
18629 (immed_wide_int_const): Use as_a <scalar_mode>.
18630 * explow.c (trunc_int_for_mode): Likewise.
18631 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
18632 (wi::shwi): Likewise.
18633 (wi::min_value): Likewise.
18634 (wi::max_value): Likewise.
18635 * dwarf2out.c (loc_descriptor): Likewise.
18636 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
18637 for CONST_WIDE_INT.
18639 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18640 Alan Hayward <alan.hayward@arm.com>
18641 David Sherwood <david.sherwood@arm.com>
18643 * tree.h (SCALAR_TYPE_MODE): New macro.
18644 * expr.c (expand_expr_addr_expr_1): Use it.
18645 (expand_expr_real_2): Likewise.
18646 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
18647 (fold_convert_const_fixed_from_int): Likewise.
18648 (fold_convert_const_fixed_from_real): Likewise.
18649 (native_encode_fixed): Likewise
18650 (native_encode_complex): Likewise
18651 (native_encode_vector): Likewise.
18652 (native_interpret_fixed): Likewise.
18653 (native_interpret_real): Likewise.
18654 (native_interpret_complex): Likewise.
18655 (native_interpret_vector): Likewise.
18656 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
18657 (simd_clone_adjust_argument_types): Likewise.
18658 (simd_clone_init_simd_arrays): Likewise.
18659 (simd_clone_adjust): Likewise.
18660 * stor-layout.c (layout_type): Likewise.
18661 * tree.c (build_minus_one_cst): Likewise.
18662 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18663 * tree-inline.c (estimate_move_cost): Likewise.
18664 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
18665 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
18666 (vectorizable_reduction): Likewise.
18667 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
18668 (vect_recog_mixed_size_cond_pattern): Likewise.
18669 (check_bool_pattern): Likewise.
18670 (adjust_bool_pattern): Likewise.
18671 (search_type_for_mask_1): Likewise.
18672 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
18673 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
18674 (vectorizable_load): Likewise.
18675 (vectorizable_store): Likewise.
18676 * ubsan.c (ubsan_encode_value): Likewise.
18677 * varasm.c (output_constant): Likewise.
18679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18680 Alan Hayward <alan.hayward@arm.com>
18681 David Sherwood <david.sherwood@arm.com>
18683 * coretypes.h (scalar_mode): New class.
18684 * machmode.h (scalar_mode): Likewise.
18685 (scalar_mode::includes_p): New function.
18686 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
18687 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
18688 * genmodes.c (get_mode_class): Handle remaining scalar modes.
18689 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
18690 * expmed.c (store_bit_field_1): Likewise.
18691 (extract_bit_field_1): Likewise.
18692 * expr.c (write_complex_part): Likewise.
18693 (read_complex_part): Likewise.
18694 (emit_move_complex_push): Likewise.
18695 (expand_expr_real_2): Likewise.
18696 * function.c (assign_parm_setup_reg): Likewise.
18697 (assign_parms_unsplit_complex): Likewise.
18698 * optabs.c (expand_binop): Likewise.
18699 * rtlanal.c (subreg_get_info): Likewise.
18700 * simplify-rtx.c (simplify_immed_subreg): Likewise.
18701 * varasm.c (output_constant_pool_2): 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 * expmed.c (extract_high_half): Use scalar_int_mode and remove
18709 (expmed_mult_highpart_optab): Likewise.
18710 (expmed_mult_highpart): Likewise.
18712 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18713 Alan Hayward <alan.hayward@arm.com>
18714 David Sherwood <david.sherwood@arm.com>
18716 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
18717 instead of a machine_mode.
18718 (builtin_memset_read_str): Likewise.
18719 * builtins.c (c_readstr): Likewise.
18720 (builtin_memcpy_read_str): Likewise.
18721 (builtin_strncpy_read_str): Likewise.
18722 (builtin_memset_read_str): Likewise.
18723 (builtin_memset_gen_str): Likewise.
18724 (expand_builtin_signbit): Use scalar_int_mode for local variables.
18725 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
18726 instead of a machine_mode.
18727 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
18729 (make_extraction): Likewise.
18730 (try_widen_shift_mode): Take and return scalar_int_modes instead
18732 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
18733 a scalar_int_mode instead of a machine_mode.
18734 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
18735 (avr_addr_space_pointer_mode): Likewise.
18736 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
18737 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
18738 (msp430_unwind_word_mode): Likewise.
18739 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
18740 (spu_addr_space_pointer_mode): Likewise.
18741 (spu_addr_space_address_mode): Likewise.
18742 (spu_libgcc_cmp_return_mode): Likewise.
18743 (spu_libgcc_shift_count_mode): Likewise.
18744 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
18745 (rl78_addr_space_pointer_mode): Likewise.
18746 (fl78_unwind_word_mode): Likewise.
18747 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
18749 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
18750 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
18751 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
18752 (mips_valid_pointer_mode): Likewise.
18753 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
18754 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
18755 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
18757 (ft32_addr_space_address_mode): Likewise.
18758 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
18759 scalar_int_mode instead of a machine_mode.
18760 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
18762 (m32c_addr_space_address_mode): Likewise.
18763 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
18764 (rs6000_eh_return_filter_mode): Likewise.
18765 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
18766 (rs6000_eh_return_filter_mode): Likewise.
18767 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
18768 (s390_libgcc_shift_count_mode): Likewise.
18769 (s390_unwind_word_mode): Likewise.
18770 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
18772 * target.def (mode_rep_extended): Likewise.
18773 (valid_pointer_mode): Likewise.
18774 (addr_space.valid_pointer_mode): Likewise.
18775 (eh_return_filter_mode): Return a scalar_int_mode rather than
18777 (libgcc_cmp_return_mode): Likewise.
18778 (libgcc_shift_count_mode): Likewise.
18779 (unwind_word_mode): Likewise.
18780 (addr_space.pointer_mode): Likewise.
18781 (addr_space.address_mode): Likewise.
18782 * doc/tm.texi: Regenerate.
18783 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
18785 (do_jump): Use scalar_int_mode for local variables.
18786 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
18787 rather than a machine_mode.
18788 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
18789 (scompare_loc_descriptor_wide): Likewise.
18790 (scompare_loc_descriptor_narrow): Likewise.
18791 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
18793 * except.c (sjlj_emit_dispatch_table): Likewise.
18794 (expand_builtin_eh_copy_values): Likewise.
18795 * explow.c (convert_memory_address_addr_space_1): Likewise.
18796 Take a scalar_int_mode rather than a machine_mode.
18797 (convert_memory_address_addr_space): Take a scalar_int_mode rather
18798 than a machine_mode.
18799 (memory_address_addr_space): Use scalar_int_mode for local variables.
18800 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
18801 rather than a machine_mode.
18802 * expmed.c (mask_rtx): Likewise.
18803 (init_expmed_one_conv): Likewise.
18804 (expand_mult_highpart_adjust): Likewise.
18805 (extract_high_half): Likewise.
18806 (expmed_mult_highpart_optab): Likewise.
18807 (expmed_mult_highpart): Likewise.
18808 (expand_smod_pow2): Likewise.
18809 (expand_sdiv_pow2): Likewise.
18810 (emit_store_flag_int): Likewise.
18811 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
18813 (extract_low_bits): Likewise.
18814 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
18816 * expr.c (pieces_addr::adjust): Likewise.
18817 (can_store_by_pieces): Likewise.
18818 (store_by_pieces): Likewise.
18819 (clear_by_pieces_1): Likewise.
18820 (expand_expr_addr_expr_1): Likewise.
18821 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
18822 (expand_expr_real_1): Likewise.
18823 (try_casesi): Likewise.
18824 * final.c (shorten_branches): Likewise.
18825 * fold-const.c (fold_convert_const_int_from_fixed): Change the
18826 type of "mode" to machine_mode.
18827 * internal-fn.c (expand_arith_overflow_result_store): Take a
18828 scalar_int_mode rather than a machine_mode.
18829 (expand_mul_overflow): Use scalar_int_mode for local variables.
18830 * loop-doloop.c (doloop_modify): Likewise.
18831 (doloop_optimize): Likewise.
18832 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
18833 than a machine_mode.
18834 (expand_doubleword_shift_condmove): Likewise.
18835 (expand_doubleword_shift): Likewise.
18836 (expand_doubleword_clz): Likewise.
18837 (expand_doubleword_popcount): Likewise.
18838 (expand_doubleword_parity): Likewise.
18839 (expand_absneg_bit): Use scalar_int_mode for local variables.
18840 (prepare_float_lib_cmp): Likewise.
18841 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
18842 rather than a machine_mode.
18843 (convert_memory_address_addr_space): Likewise.
18844 (get_mode_bounds): Likewise.
18845 (get_address_mode): Return a scalar_int_mode rather than a
18847 * rtlanal.c (get_address_mode): Likewise.
18848 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
18849 than a machine_mode.
18850 * targhooks.c (default_mode_rep_extended): Likewise.
18851 (default_valid_pointer_mode): Likewise.
18852 (default_addr_space_valid_pointer_mode): Likewise.
18853 (default_eh_return_filter_mode): Return a scalar_int_mode rather
18854 than a machine_mode.
18855 (default_libgcc_cmp_return_mode): Likewise.
18856 (default_libgcc_shift_count_mode): Likewise.
18857 (default_unwind_word_mode): Likewise.
18858 (default_addr_space_pointer_mode): Likewise.
18859 (default_addr_space_address_mode): Likewise.
18860 * targhooks.h (default_eh_return_filter_mode): Likewise.
18861 (default_libgcc_cmp_return_mode): Likewise.
18862 (default_libgcc_shift_count_mode): Likewise.
18863 (default_unwind_word_mode): Likewise.
18864 (default_addr_space_pointer_mode): Likewise.
18865 (default_addr_space_address_mode): Likewise.
18866 (default_mode_rep_extended): Take a scalar_int_mode rather than
18868 (default_valid_pointer_mode): Likewise.
18869 (default_addr_space_valid_pointer_mode): Likewise.
18870 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
18872 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
18873 rather than a machine_mode.
18874 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
18875 for local variables.
18876 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
18877 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
18878 than a machine_mode.
18880 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18881 Alan Hayward <alan.hayward@arm.com>
18882 David Sherwood <david.sherwood@arm.com>
18884 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
18885 the mode argument to scalar_int_mode.
18886 (do_jump_by_parts_zero_rtx): Likewise.
18887 (do_jump_by_parts_equality_rtx): Likewise.
18888 (do_jump_by_parts_greater): Take a mode argument.
18889 (do_jump_by_parts_equality): Likewise.
18890 (do_jump_1): Update calls accordingly.
18892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18893 Alan Hayward <alan.hayward@arm.com>
18894 David Sherwood <david.sherwood@arm.com>
18896 * is-a.h (safe_dyn_cast): New function.
18897 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
18898 (jump_table_for_label): Likewise.
18899 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
18900 instead of an rtx_insn *.
18901 (shorten_branches): Use dyn_cast instead of LABEL_P and
18902 JUMP_TABLE_DATA_P. Use jump_table_for_label and
18903 rtx_jump_table_data::get_data_mode.
18904 (final_scan_insn): Likewise.
18906 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18907 Alan Hayward <alan.hayward@arm.com>
18908 David Sherwood <david.sherwood@arm.com>
18910 * combine.c (try_combine): Use is_a <scalar_int_mode> when
18911 trying to combine a full-register integer set with a subreg
18914 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18915 Alan Hayward <alan.hayward@arm.com>
18916 David Sherwood <david.sherwood@arm.com>
18918 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
18919 that is always either address_mode or pointer_mode.
18921 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18922 Alan Hayward <alan.hayward@arm.com>
18923 David Sherwood <david.sherwood@arm.com>
18925 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
18926 when the two are known to be equal.
18928 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18929 Alan Hayward <alan.hayward@arm.com>
18930 David Sherwood <david.sherwood@arm.com>
18932 * simplify-rtx.c (simplify_const_unary_operation): Use
18933 is_a <scalar_int_mode> instead of checking for a nonzero
18934 precision. Forcibly convert op_mode to a scalar_int_mode
18935 in that case. More clearly differentiate the operand and
18936 result modes and use the former when deciding what the value
18937 of a count-bits operation should be. Use is_int_mode instead
18938 of checking for a MODE_INT. Remove redundant check for whether
18939 this mode has a zero precision.
18941 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18942 Alan Hayward <alan.hayward@arm.com>
18943 David Sherwood <david.sherwood@arm.com>
18945 * optabs.c (widen_leading): Change the type of the mode argument
18946 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
18947 (widen_bswap): Likewise.
18948 (expand_parity): Likewise.
18949 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
18950 (expand_ffs): Likewise.
18951 (epand_unop): Check for scalar integer modes before calling the
18954 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18955 Alan Hayward <alan.hayward@arm.com>
18956 David Sherwood <david.sherwood@arm.com>
18958 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
18960 (expand_expr_real_1): Update call accordingly.
18962 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18963 Alan Hayward <alan.hayward@arm.com>
18964 David Sherwood <david.sherwood@arm.com>
18966 * expmed.c (store_bit_field_using_insv): Add op0_mode and
18967 value_mode arguments. Use scalar_int_mode internally.
18968 (store_bit_field_1): Rename the new integer mode from imode
18969 to op0_mode and use it instead of GET_MODE (op0). Update calls
18970 to store_split_bit_field, store_bit_field_using_insv and
18971 store_fixed_bit_field.
18972 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
18973 Use scalar_int_mode internally. Use a bit count rather than a mode
18974 when calculating the largest bit size for get_best_mode.
18975 Update calls to store_split_bit_field and store_fixed_bit_field_1.
18976 (store_fixed_bit_field_1): Add mode and value_mode arguments.
18977 Remove assertion that OP0 has a scalar integer mode.
18978 (store_split_bit_field): Add op0_mode and value_mode arguments.
18979 Update calls to extract_fixed_bit_field.
18980 (extract_bit_field_using_extv): Add an op0_mode argument.
18981 Use scalar_int_mode internally.
18982 (extract_bit_field_1): Rename the new integer mode from imode to
18983 op0_mode and use it instead of GET_MODE (op0). Update calls to
18984 extract_split_bit_field, extract_bit_field_using_extv and
18985 extract_fixed_bit_field.
18986 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
18987 to extract_split_bit_field and extract_fixed_bit_field_1.
18988 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
18989 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
18990 on the target mode.
18991 (extract_split_bit_field): Add an op0_mode argument. Update call
18992 to extract_fixed_bit_field.
18994 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
18995 Alan Hayward <alan.hayward@arm.com>
18996 David Sherwood <david.sherwood@arm.com>
18998 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
18999 * explow.c (hard_function_value): Likewise.
19000 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
19001 convert_to_mode call outside the loop.
19002 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
19003 for the mode iterator. Require the mode specified by max_pieces
19005 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
19007 (copy_blkmode_to_reg): Likewise.
19008 (set_storage_via_setmem): Likewise.
19009 * optabs.c (prepare_cmp_insn): Likewise.
19010 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19011 * stor-layout.c (finish_bitfield_representative): 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 (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
19018 * expr.c (convert_move): Use them.
19019 (convert_modes): Likewise.
19020 (store_expr_with_bounds): Likewise.
19022 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19023 Alan Hayward <alan.hayward@arm.com>
19024 David Sherwood <david.sherwood@arm.com>
19026 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
19027 parameter for the mode of "x". Remove the "known_x", "known_mode"
19028 and "known_ret" arguments. Change the type of the mode argument
19029 to scalar_int_mode.
19030 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
19031 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
19032 (reg_num_sign_bit_copies_for_combine): Likewise.
19033 * rtlanal.c (nonzero_bits1): Likewise.
19034 (num_sign_bit_copies1): Likewise.
19035 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
19036 (reg_num_sign_bit_copies_general): Likewise.
19037 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
19038 (reg_nonzero_bits_general): Likewise.
19040 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19041 Alan Hayward <alan.hayward@arm.com>
19042 David Sherwood <david.sherwood@arm.com>
19044 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
19045 than in subroutines. Return 1 for non-integer modes.
19046 (cached_num_sign_bit_copies): Change the type of the mode parameter
19047 to scalar_int_mode.
19048 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
19049 classes. Handle CONST_INT_P first and then check whether X also
19050 has a scalar integer mode. Check the same thing for inner registers
19051 of a SUBREG and for values that are being extended or truncated.
19053 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19054 Alan Hayward <alan.hayward@arm.com>
19055 David Sherwood <david.sherwood@arm.com>
19057 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
19058 in subroutines. Return the mode mask for non-integer modes.
19059 (cached_nonzero_bits): Change the type of the mode parameter
19060 to scalar_int_mode.
19061 (nonzero_bits1): Likewise. Remove early exit for other mode
19062 classes. Handle CONST_INT_P first and then check whether X
19063 also has a scalar integer mode.
19065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19066 Alan Hayward <alan.hayward@arm.com>
19067 David Sherwood <david.sherwood@arm.com>
19069 * expr.c (widest_int_mode_for_size): Make the comment match the code.
19070 Return a scalar_int_mode and assert that the size is greater than
19072 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
19073 (op_by_pieces_d::op_by_pieces_d): Likewise.
19074 (op_by_pieces_d::run): Likewise.
19075 (can_store_by_pieces): Likewise.
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 (extract_left_shift): Add a mode argument and update
19083 (make_compound_operation_int): Change the type of the mode parameter
19084 to scalar_int_mode and update the call to extract_left_shift.
19086 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19087 Alan Hayward <alan.hayward@arm.com>
19088 David Sherwood <david.sherwood@arm.com>
19090 * combine.c (simplify_and_const_int): Change the type of the mode
19091 parameter to scalar_int_mode.
19092 (simplify_and_const_int_1): Likewise. Update recursive call.
19094 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19095 Alan Hayward <alan.hayward@arm.com>
19096 David Sherwood <david.sherwood@arm.com>
19098 * combine.c (simplify_compare_const): Check that the mode is a
19099 scalar_int_mode (rather than VOIDmode) before testing its
19101 (simplify_comparison): Move COMPARISON_P handling out of the
19102 loop and restrict the latter part of the loop to scalar_int_modes.
19103 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
19104 and when considering SUBREG_REGs. Use is_int_mode instead of
19105 checking GET_MODE_CLASS against MODE_INT.
19107 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19108 Alan Hayward <alan.hayward@arm.com>
19109 David Sherwood <david.sherwood@arm.com>
19111 * combine.c (try_widen_shift_mode): Move check for equal modes to...
19112 (simplify_shift_const_1): ...here. Use scalar_int_mode for
19113 shift_unit_mode and for modes involved in scalar shifts.
19115 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19116 Alan Hayward <alan.hayward@arm.com>
19117 David Sherwood <david.sherwood@arm.com>
19119 * combine.c (force_int_to_mode): New function, split out from...
19120 (force_to_mode): ...here. Keep xmode up-to-date and use it
19121 instead of GET_MODE (x).
19123 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19124 Alan Hayward <alan.hayward@arm.com>
19125 David Sherwood <david.sherwood@arm.com>
19127 * optabs-query.h (extraction_insn::struct_mode): Change type to
19128 opt_scalar_int_mode and update comment.
19129 (extraction_insn::field_mode): Change type to scalar_int_mode.
19130 (extraction_insn::pos_mode): Likewise.
19131 * combine.c (make_extraction): Update accordingly.
19132 * optabs-query.c (get_traditional_extraction_insn): Likewise.
19133 (get_optab_extraction_insn): Likewise.
19134 * recog.c (simplify_while_replacing): Likewise.
19135 * expmed.c (narrow_bit_field_mem): Change the type of the mode
19136 parameter to opt_scalar_int_mode.
19138 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19139 Alan Hayward <alan.hayward@arm.com>
19140 David Sherwood <david.sherwood@arm.com>
19142 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
19143 to a scalar_int_mode instead of a machine_mode.
19144 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
19145 (get_best_mode): Return a boolean and use a pointer argument to store
19146 the selected mode. Replace the limit mode parameter with a bit limit.
19147 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
19148 for the values returned by bit_field_mode_iterator::next_mode.
19149 (store_bit_field): Update call to get_best_mode.
19150 (store_fixed_bit_field): Likewise.
19151 (extract_fixed_bit_field): Likewise.
19152 * expr.c (optimize_bitfield_assignment_op): Likewise.
19153 * fold-const.c (optimize_bit_field_compare): Likewise.
19154 (fold_truth_andor_1): Likewise.
19155 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
19156 Update for new type of m_mode.
19157 (get_best_mode): As above.
19159 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19160 Alan Hayward <alan.hayward@arm.com>
19161 David Sherwood <david.sherwood@arm.com>
19163 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
19164 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
19165 (store_bit_field): Check is_a <scalar_int_mode> before calling
19166 strict_volatile_bitfield_p.
19167 (extract_bit_field): Likewise.
19169 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19170 Alan Hayward <alan.hayward@arm.com>
19171 David Sherwood <david.sherwood@arm.com>
19173 * target.def (cstore_mode): Return a scalar_int_mode.
19174 * doc/tm.texi: Regenerate.
19175 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
19176 * targhooks.h (default_cstore_mode): Likewise.
19177 * targhooks.c (default_cstore_mode): Likewise, using a forced
19179 * expmed.c (emit_cstore): Expect the target of the cstore to be
19182 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19183 Alan Hayward <alan.hayward@arm.com>
19184 David Sherwood <david.sherwood@arm.com>
19186 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
19188 (niter_desc): Likewise mode.
19189 (iv_analyze): Add a mode parameter.
19191 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
19192 and change its type to scalar_int_mode.
19193 * loop-iv.c: Update commentary at head of file.
19194 (iv_constant): Pass the mode paraeter before the rtx it describes
19195 and change its type to scalar_int_mode. Remove VOIDmode handling.
19196 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
19197 (iv_extend): Likewise.
19198 (shorten_into_mode): Likewise.
19199 (iv_add): Use scalar_int_mode.
19200 (iv_mult): Likewise.
19201 (iv_shift): Likewise.
19202 (canonicalize_iv_subregs): Likewise.
19203 (get_biv_step_1): Pass the outer_mode parameter before the rtx
19204 it describes and change its mode to scalar_int_mode. Also change
19205 the type of the returned inner_mode to scalar_int_mode.
19206 (get_biv_step): Likewise, turning outer_mode from a pointer
19207 into a direct parameter. Update call to get_biv_step_1.
19208 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
19209 iv_constant and get_biv_step.
19210 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
19211 and change its type to scalar_int_mode. Don't initialise iv->mode
19212 to VOIDmode and remove later checks for its still being VOIDmode.
19213 Update calls to iv_analyze_op and iv_analyze_expr. Check
19214 is_a <scalar_int_mode> when changing the mode under consideration.
19215 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
19216 Update call to iv_analyze_expr.
19217 (iv_analyze_op): Add a mode parameter. Reject subregs whose
19218 inner register is not also a scalar_int_mode. Update call to
19220 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
19221 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
19222 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
19223 separate mode class checks. Update calls to iv_analyze. Remove
19224 fix-up of VOIDmodes after iv_analyze_biv.
19225 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
19226 don't have a scalar_int_mode. Update call to biv_p.
19228 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19229 Alan Hayward <alan.hayward@arm.com>
19230 David Sherwood <david.sherwood@arm.com>
19232 * cfgexpand.c (convert_debug_memory_address): Use
19233 as_a <scalar_int_mode>.
19234 * combine.c (expand_compound_operation): Likewise.
19235 (make_extraction): Likewise.
19236 (change_zero_ext): Likewise.
19237 (simplify_comparison): Likewise.
19238 * cse.c (cse_insn): Likewise.
19239 * dwarf2out.c (minmax_loc_descriptor): Likewise.
19240 (mem_loc_descriptor): Likewise.
19241 (loc_descriptor): Likewise.
19242 * expmed.c (init_expmed_one_mode): Likewise.
19243 (synth_mult): Likewise.
19244 (emit_store_flag_1): Likewise.
19245 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
19246 of a comparison with size.
19247 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
19248 (reduce_to_bit_field_precision): Likewise.
19249 * function.c (expand_function_end): Likewise.
19250 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
19251 * loop-doloop.c (doloop_modify): Likewise.
19252 * optabs.c (expand_binop): Likewise.
19253 (expand_unop): Likewise.
19254 (expand_copysign_absneg): Likewise.
19255 (prepare_cmp_insn): Likewise.
19256 (maybe_legitimize_operand): Likewise.
19257 * recog.c (const_scalar_int_operand): Likewise.
19258 * rtlanal.c (get_address_mode): Likewise.
19259 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19260 (simplify_cond_clz_ctz): Likewise.
19261 * tree-nested.c (get_nl_goto_field): Likewise.
19262 * tree.c (build_vector_type_for_mode): Likewise.
19263 * var-tracking.c (use_narrower_mode): Likewise.
19265 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19266 Alan Hayward <alan.hayward@arm.com>
19267 David Sherwood <david.sherwood@arm.com>
19269 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
19270 * builtins.c (expand_builtin_signbit): Use it.
19271 * cfgexpand.c (expand_debug_expr): Likewise.
19272 * dojump.c (do_jump): Likewise.
19273 (do_compare_and_jump): Likewise.
19274 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
19275 * expmed.c (make_tree): Likewise.
19276 * expr.c (expand_expr_real_2): Likewise.
19277 (expand_expr_real_1): Likewise.
19278 (try_casesi): Likewise.
19279 * fold-const-call.c (fold_const_call_ss): Likewise.
19280 * fold-const.c (unextend): Likewise.
19281 (extract_muldiv_1): Likewise.
19282 (fold_single_bit_test): Likewise.
19283 (native_encode_int): Likewise.
19284 (native_encode_string): Likewise.
19285 (native_interpret_int): Likewise.
19286 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
19287 * internal-fn.c (expand_addsub_overflow): Likewise.
19288 (expand_neg_overflow): Likewise.
19289 (expand_mul_overflow): Likewise.
19290 (expand_arith_overflow): Likewise.
19291 * match.pd: Likewise.
19292 * stor-layout.c (layout_type): Likewise.
19293 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
19294 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
19295 * tree-ssanames.c (get_range_info): Likewise.
19296 * tree-switch-conversion.c (array_value_type) Likewise.
19297 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
19298 (vect_recog_divmod_pattern): Likewise.
19299 (vect_recog_mixed_size_cond_pattern): Likewise.
19300 * tree-vrp.c (extract_range_basic): Likewise.
19301 (simplify_float_conversion_using_ranges): Likewise.
19302 * tree.c (int_fits_type_p): Likewise.
19303 * ubsan.c (instrument_bool_enum_load): Likewise.
19304 * varasm.c (mergeable_string_section): Likewise.
19305 (narrowing_initializer_constant_valid_p): Likewise.
19306 (output_constant): Likewise.
19308 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19309 Alan Hayward <alan.hayward@arm.com>
19310 David Sherwood <david.sherwood@arm.com>
19312 * machmode.h (NARROWEST_INT_MODE): New macro.
19313 * expr.c (alignment_for_piecewise_move): Use it instead of
19314 GET_CLASS_NARROWEST_MODE (MODE_INT).
19315 (push_block): Likewise.
19316 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
19318 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19320 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19321 Alan Hayward <alan.hayward@arm.com>
19322 David Sherwood <david.sherwood@arm.com>
19324 * postreload.c (move2add_valid_value_p): Change the type of the
19325 mode parameter to scalar_int_mode.
19326 (move2add_use_add2_insn): Add a mode parameter and use it instead
19328 (move2add_use_add3_insn): Likewise.
19329 (reload_cse_move2add): Update accordingly.
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 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
19337 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
19338 * optabs.c (expand_unop): Likewise.
19340 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19341 Alan Hayward <alan.hayward@arm.com>
19342 David Sherwood <david.sherwood@arm.com>
19344 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
19345 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
19346 (popcount_loc_descriptor): Likewise.
19347 (bswap_loc_descriptor): Likewise.
19348 (rotate_loc_descriptor): Likewise.
19349 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
19350 calling the functions above.
19352 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19353 Alan Hayward <alan.hayward@arm.com>
19354 David Sherwood <david.sherwood@arm.com>
19356 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
19358 (try_combine): Likewise.
19359 (simplify_if_then_else): Likewise.
19360 * cse.c (cse_insn): Likewise.
19361 * dwarf2out.c (mem_loc_descriptor): Likewise.
19362 * emit-rtl.c (gen_lowpart_common): Likewise.
19363 * simplify-rtx.c (simplify_truncation): Likewise.
19364 (simplify_binary_operation_1): Likewise.
19365 (simplify_const_relational_operation): Likewise.
19366 (simplify_ternary_operation): Likewise.
19367 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
19369 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19370 Alan Hayward <alan.hayward@arm.com>
19371 David Sherwood <david.sherwood@arm.com>
19373 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
19374 * reload.c (push_reload): Likewise.
19375 (find_reloads): Likewise.
19377 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19378 Alan Hayward <alan.hayward@arm.com>
19379 David Sherwood <david.sherwood@arm.com>
19381 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
19382 (make_compound_operation_int): Likewise.
19383 (change_zero_ext): Likewise.
19384 * expr.c (convert_move): Likewise.
19385 (convert_modes): Likewise.
19386 * fwprop.c (forward_propagate_subreg): Likewise.
19387 * loop-iv.c (get_biv_step_1): Likewise.
19388 * optabs.c (widen_operand): Likewise.
19389 * postreload.c (move2add_valid_value_p): Likewise.
19390 * recog.c (simplify_while_replacing): Likewise.
19391 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19392 (simplify_binary_operation_1): Likewise. Remove redundant
19393 mode equality check.
19395 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19396 Alan Hayward <alan.hayward@arm.com>
19397 David Sherwood <david.sherwood@arm.com>
19399 * combine.c (combine_simplify_rtx): Add checks for
19400 is_a <scalar_int_mode>.
19401 (simplify_if_then_else): Likewise.
19402 (make_field_assignment): Likewise.
19403 (simplify_comparison): Likewise.
19404 * ifcvt.c (noce_try_bitop): Likewise.
19405 * loop-invariant.c (canonicalize_address_mult): Likewise.
19406 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19409 Alan Hayward <alan.hayward@arm.com>
19410 David Sherwood <david.sherwood@arm.com>
19412 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
19413 is_a <scalar_int_mode> instead of != BLKmode.
19415 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19416 Alan Hayward <alan.hayward@arm.com>
19417 David Sherwood <david.sherwood@arm.com>
19419 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
19420 instead of != VOIDmode.
19421 * combine.c (if_then_else_cond): Likewise.
19422 (change_zero_ext): Likewise.
19423 * dwarf2out.c (mem_loc_descriptor): Likewise.
19424 (loc_descriptor): Likewise.
19425 * rtlanal.c (canonicalize_condition): Likewise.
19426 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
19428 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19429 Alan Hayward <alan.hayward@arm.com>
19430 David Sherwood <david.sherwood@arm.com>
19432 * simplify-rtx.c (simplify_binary_operation_1): Use
19433 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
19435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19436 Alan Hayward <alan.hayward@arm.com>
19437 David Sherwood <david.sherwood@arm.com>
19439 * wide-int.h (int_traits<unsigned char>) New class.
19440 (int_traits<unsigned short>) Likewise.
19441 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
19442 Use GET_MODE_UNIT_PRECISION and remove redundant test for
19444 * combine.c (set_nonzero_bits_and_sign_copies): Use
19445 is_a <scalar_int_mode>.
19446 (find_split_point): Likewise.
19447 (combine_simplify_rtx): Likewise.
19448 (simplify_logical): Likewise.
19449 (expand_compound_operation): Likewise.
19450 (expand_field_assignment): Likewise.
19451 (make_compound_operation): Likewise.
19452 (extended_count): Likewise.
19453 (change_zero_ext): Likewise.
19454 (simplify_comparison): Likewise.
19455 * dwarf2out.c (scompare_loc_descriptor): Likewise.
19456 (ucompare_loc_descriptor): Likewise.
19457 (minmax_loc_descriptor): Likewise.
19458 (mem_loc_descriptor): Likewise.
19459 (loc_descriptor): Likewise.
19460 * expmed.c (init_expmed_one_mode): Likewise.
19461 * lra-constraints.c (lra_constraint_offset): Likewise.
19462 * optabs.c (prepare_libcall_arg): Likewise.
19463 * postreload.c (move2add_note_store): Likewise.
19464 * reload.c (operands_match_p): Likewise.
19465 * rtl.h (load_extend_op): Likewise.
19466 * rtlhooks.c (gen_lowpart_general): Likewise.
19467 * simplify-rtx.c (simplify_truncation): Likewise.
19468 (simplify_unary_operation_1): Likewise.
19469 (simplify_binary_operation_1): Likewise.
19470 (simplify_const_binary_operation): Likewise.
19471 (simplify_const_relational_operation): Likewise.
19472 (simplify_subreg): Likewise.
19473 * stor-layout.c (bitwise_mode_for_mode): Likewise.
19474 * var-tracking.c (adjust_mems): Likewise.
19475 (prepare_call_arguments): Likewise.
19477 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19478 Alan Hayward <alan.hayward@arm.com>
19479 David Sherwood <david.sherwood@arm.com>
19481 * machmode.h (is_int_mode): New fuction.
19482 * combine.c (find_split_point): Use it.
19483 (combine_simplify_rtx): Likewise.
19484 (simplify_if_then_else): Likewise.
19485 (simplify_set): Likewise.
19486 (simplify_shift_const_1): Likewise.
19487 (simplify_comparison): Likewise.
19488 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
19489 * cse.c (notreg_cost): Likewise.
19490 (cse_insn): Likewise.
19491 * cselib.c (cselib_lookup_1): Likewise.
19492 * dojump.c (do_jump_1): Likewise.
19493 (do_compare_rtx_and_jump): Likewise.
19494 * dse.c (get_call_args): Likewise.
19495 * dwarf2out.c (rtl_for_decl_init): Likewise.
19496 (native_encode_initializer): Likewise.
19497 * expmed.c (emit_store_flag_1): Likewise.
19498 (emit_store_flag): Likewise.
19499 * expr.c (convert_modes): Likewise.
19500 (store_field): Likewise.
19501 (expand_expr_real_1): Likewise.
19502 * fold-const.c (fold_read_from_constant_string): Likewise.
19503 * gimple-ssa-sprintf.c (get_format_string): Likewise.
19504 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
19505 * optabs.c (expand_binop): Likewise.
19506 (expand_unop): Likewise.
19507 (expand_abs_nojump): Likewise.
19508 (expand_one_cmpl_abs_nojump): Likewise.
19509 * simplify-rtx.c (mode_signbit_p): Likewise.
19510 (val_signbit_p): Likewise.
19511 (val_signbit_known_set_p): Likewise.
19512 (val_signbit_known_clear_p): Likewise.
19513 (simplify_relational_operation_1): Likewise.
19514 * tree.c (vector_type_mode): Likewise.
19516 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19517 Alan Hayward <alan.hayward@arm.com>
19518 David Sherwood <david.sherwood@arm.com>
19520 * machmode.h (smallest_mode_for_size): Fix formatting.
19521 (smallest_int_mode_for_size): New function.
19522 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
19523 instead of smallest_mode_for_size.
19524 * combine.c (make_extraction): Likewise.
19525 * config/arc/arc.c (arc_expand_movmem): Likewise.
19526 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
19527 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
19528 * config/s390/s390.c (s390_expand_insv): Likewise.
19529 * config/sparc/sparc.c (assign_int_registers): Likewise.
19530 * config/spu/spu.c (spu_function_value): Likewise.
19531 (spu_function_arg): Likewise.
19532 * coverage.c (get_gcov_type): Likewise.
19533 (get_gcov_unsigned_t): Likewise.
19534 * dse.c (find_shift_sequence): Likewise.
19535 * expmed.c (store_bit_field_1): Likewise.
19536 * expr.c (convert_move): Likewise.
19537 (store_field): Likewise.
19538 * internal-fn.c (expand_arith_overflow): Likewise.
19539 * optabs-query.c (get_best_extraction_insn): Likewise.
19540 * optabs.c (expand_twoval_binop_libfunc): Likewise.
19541 * stor-layout.c (layout_type): Likewise.
19542 (initialize_sizetypes): Likewise.
19543 * targhooks.c (default_get_mask_mode): Likewise.
19544 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
19546 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19547 Alan Hayward <alan.hayward@arm.com>
19548 David Sherwood <david.sherwood@arm.com>
19550 * machmode.h (opt_mode::else_blk): New function.
19551 (int_mode_for_mode): Declare.
19552 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
19553 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
19555 * cfgexpand.c (expand_debug_expr): Likewise.
19556 * combine.c (gen_lowpart_or_truncate): Likewise.
19557 (gen_lowpart_for_combine): Likewise.
19558 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
19559 * config/avr/avr.c (avr_to_int_mode): Likewise.
19560 (avr_out_plus_1): Likewise.
19561 (avr_out_plus): Likewise.
19562 (avr_out_round): Likewise.
19563 * config/i386/i386.c (ix86_split_to_parts): Likewise.
19564 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
19565 (s390_expand_vcond): Likewise.
19566 * config/spu/spu.c (spu_split_immediate): Likewise.
19567 (spu_expand_mov): Likewise.
19568 * dse.c (get_stored_val): Likewise.
19569 * expmed.c (store_bit_field_1): Likewise.
19570 (convert_extracted_bit_field): Use int_mode_for_mode instead of
19572 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
19573 (extract_low_bits): Likewise.
19574 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
19575 handling rather than repeating the check.
19576 (emit_group_store): Likewise.
19577 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
19578 * optabs.c (expand_absneg_bit): Likewise.
19579 (expand_copysign_absneg): Likewise.
19580 (expand_copysign_bit): Likewise.
19581 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
19582 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
19583 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
19584 * var-tracking.c (prepare_call_arguments): Likewise.
19585 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
19586 int_mode_for_mode instead of mode_for_size.
19587 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
19589 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19590 Alan Hayward <alan.hayward@arm.com>
19591 David Sherwood <david.sherwood@arm.com>
19593 * machmode.h (int_mode_for_size): New function.
19594 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
19595 instead of mode_for_size.
19596 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
19598 * combine.c (expand_field_assignment): Use int_mode_for_size
19599 instead of mode_for_size.
19600 (make_extraction): Likewise.
19601 (simplify_shift_const_1): Likewise.
19602 (simplify_comparison): Likewise.
19603 * dojump.c (do_jump): Likewise.
19604 * dwarf2out.c (mem_loc_descriptor): Likewise.
19605 * emit-rtl.c (init_derived_machine_modes): Likewise.
19606 * expmed.c (flip_storage_order): Likewise.
19607 (convert_extracted_bit_field): Likewise.
19608 * expr.c (copy_blkmode_from_reg): Likewise.
19609 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
19610 * internal-fn.c (expand_mul_overflow): Likewise.
19611 * lower-subreg.c (simple_move): Likewise.
19612 * optabs-libfuncs.c (init_optabs): Likewise.
19613 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
19614 * tree.c (vector_type_mode): Likewise.
19615 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
19616 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
19617 * tree-vect-generic.c (expand_vector_parallel): Likewise.
19618 * tree-vect-stmts.c (vectorizable_load): Likewise.
19619 (vectorizable_store): Likewise.
19621 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19622 Alan Hayward <alan.hayward@arm.com>
19623 David Sherwood <david.sherwood@arm.com>
19625 * coretypes.h (pod_mode): New type.
19626 (scalar_int_mode_pod): New typedef.
19627 * machmode.h (pod_mode): New class.
19628 (int_n_data_t::m): Change type to scalar_int_mode_pod.
19629 * genmodes.c (emit_mode_int_n): Update accordingly.
19630 * lower-subreg.h (target_lower_subreg): Change type to
19631 scalar_int_mode_pod.
19632 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
19633 scalar_int_mode_pod.
19635 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19636 Alan Hayward <alan.hayward@arm.com>
19637 David Sherwood <david.sherwood@arm.com>
19639 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
19640 machine_mode to scalar_int_mode.
19641 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
19642 (rs6000_option_override_internal): Remove cast to int.
19643 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
19644 machine_mode to scalar_int_mode.
19645 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
19646 (rs6000_option_override_internal): Remove cast to int.
19647 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
19648 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
19650 * config/s390/s390.c (s390_expand_builtin): Likewise.
19651 * coretypes.h (scalar_int_mode): New type.
19652 (opt_scalar_int_mode): New typedef.
19653 * machmode.h (scalar_int_mode): New class.
19654 (scalar_int_mode::includes_p): New function.
19655 (byte_mode): Change type to scalar_int_mode.
19656 (word_mode): Likewise.
19657 (ptr_mode): Likewise.
19658 * emit-rtl.c (byte_mode): Likewise.
19659 (word_mode): Likewise.
19660 (ptr_mode): Likewise.
19661 (init_derived_machine_modes): Update accordingly.
19662 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
19663 and MODE_PARTIAL_INT.
19664 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
19665 opt_scalar_int_mode.
19667 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19668 Alan Hayward <alan.hayward@arm.com>
19669 David Sherwood <david.sherwood@arm.com>
19671 * target.def (libgcc_floating_mode_supported_p): Take a
19673 * doc/tm.texi: Regenerate.
19674 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
19676 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
19677 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
19680 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19681 Alan Hayward <alan.hayward@arm.com>
19682 David Sherwood <david.sherwood@arm.com>
19684 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
19685 * doc/tm.texi: Regenerate.
19686 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
19687 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
19688 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
19689 * targhooks.h (default_floatn_mode): Likewise.
19690 * targhooks.c (default_floatn_mode): Likewise.
19691 * tree.c (build_common_tree_nodes): Update accordingly.
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 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
19698 (mode_iterator::iterate_p): Likewise.
19699 (mode_iterator::get_wider): Likewise.
19700 * expr.c (init_expr_target): Use opt_scalar_float_mode.
19702 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19703 Alan Hayward <alan.hayward@arm.com>
19704 David Sherwood <david.sherwood@arm.com>
19706 * coretypes.h (opt_scalar_float_mode): New typedef.
19707 * machmode.h (float_mode_for_size): New function.
19708 * emit-rtl.c (double_mode): Delete.
19709 (init_emit_once): Use float_mode_for_size.
19710 * stor-layout.c (layout_type): Likewise.
19711 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
19713 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19714 Alan Hayward <alan.hayward@arm.com>
19715 David Sherwood <david.sherwood@arm.com>
19717 * output.h (assemble_real): Take a scalar_float_mode.
19718 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
19719 * config/arm/arm.md (consttable_4): Likewise.
19720 (consttable_8): Likewise.
19721 (consttable_16): Likewise.
19722 * config/mips/mips.md (consttable_float): Likewise.
19723 * config/s390/s390.c (s390_output_pool_entry): Likewise.
19724 * varasm.c (assemble_real): Take a scalar_float_mode.
19725 (output_constant_pool_2): Update accordingly.
19726 (output_constant): Likewise.
19728 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19729 Alan Hayward <alan.hayward@arm.com>
19730 David Sherwood <david.sherwood@arm.com>
19732 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
19733 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
19734 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
19735 (native_encode_real): Likewise.
19736 (native_interpret_real): Likewise.
19737 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
19738 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19740 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19741 Alan Hayward <alan.hayward@arm.com>
19742 David Sherwood <david.sherwood@arm.com>
19744 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
19745 <scalar_float_mode>. Simplify.
19746 (gen_extend_conv_libfunc): Likewise.
19748 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19749 Alan Hayward <alan.hayward@arm.com>
19750 David Sherwood <david.sherwood@arm.com>
19752 * coretypes.h (scalar_float_mode): New type.
19753 * machmode.h (mode_traits::from_int): Use machine_mode if
19754 USE_ENUM_MODES is defined.
19755 (is_a): New function.
19757 (dyn_cast): Likewise.
19758 (scalar_float_mode): New class.
19759 (scalar_float_mode::includes_p): New function.
19760 (is_float_mode): Likewise.
19761 * gdbhooks.py (MachineModePrinter): New class.
19762 (build_pretty_printer): Use it for scalar_float_mode.
19763 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
19764 (format_helper::format_helper): Turn into a template.
19765 * genmodes.c (get_mode_class): New function.
19766 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
19767 or machine_mode if none.
19768 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
19769 as_a <scalar_float_mode>.
19770 * dwarf2out.c (mem_loc_descriptor): Likewise.
19771 (insert_float): Likewise.
19772 (add_const_value_attribute): Likewise.
19773 * simplify-rtx.c (simplify_immed_subreg): Likewise.
19774 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
19775 (expand_unop): Update accordingly.
19776 (expand_abs_nojump): Likewise.
19777 (expand_copysign_absneg): Take a scalar_float_mode.
19778 (expand_copysign_bit): Likewise.
19779 (expand_copysign): Update accordingly.
19781 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19782 Alan Hayward <alan.hayward@arm.com>
19783 David Sherwood <david.sherwood@arm.com>
19785 * coretypes.h (opt_mode): New class.
19786 * machmode.h (opt_mode): Likewise.
19787 (opt_mode::else_void): New function.
19788 (opt_mode::require): Likewise.
19789 (opt_mode::exists): Likewise.
19790 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
19791 (GET_MODE_2XWIDER_MODE): Likewise.
19792 (mode_iterator::get_wider): Update accordingly.
19793 (mode_iterator::get_2xwider): Likewise.
19794 (mode_iterator::get_known_wider): Likewise, turning into a template.
19795 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
19796 forcing a wider mode to exist.
19797 * config/cr16/cr16.h (LONG_REG_P): Likewise.
19798 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19799 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
19800 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
19801 * lower-subreg.c (init_lower_subreg): Likewise.
19802 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
19803 on the final iteration.
19804 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
19805 a wider mode exists before asking for a move pattern.
19806 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
19807 forcing a wider mode to exist.
19808 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
19809 returning false if no such mode exists.
19810 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
19811 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
19812 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
19813 Avoid checking for a MODE_INT if we already know the mode is not a
19815 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
19816 forcing a wider mode to exist.
19817 (expmed_mult_highpart_optab): Likewise.
19818 (expmed_mult_highpart): Likewise.
19819 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
19821 * lto-streamer-in.c (lto_input_mode_table): Likewise.
19822 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
19823 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
19824 * internal-fn.c (expand_mul_overflow): Update use of
19825 GET_MODE_2XWIDER_MODE.
19826 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19827 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
19828 GET_MODE_WIDER_MODE.
19829 (convert_plusminus_to_widen): Likewise.
19830 * tree-switch-conversion.c (array_value_type): Likewise.
19831 * var-tracking.c (emit_note_insn_var_location): Likewise.
19832 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
19833 Return false inside rather than outside the loop if no wider mode
19835 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
19836 and GET_MODE_2XWIDER_MODE
19837 (can_compare_p): Use else_void.
19838 * gdbhooks.py (OptMachineModePrinter): New class.
19839 (build_pretty_printer): Use it for opt_mode.
19841 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19842 Alan Hayward <alan.hayward@arm.com>
19843 David Sherwood <david.sherwood@arm.com>
19845 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
19846 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
19848 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19849 Alan Hayward <alan.hayward@arm.com>
19850 David Sherwood <david.sherwood@arm.com>
19852 * machmode.h (mode_traits): New structure.
19853 (get_narrowest_mode): New function.
19854 (mode_iterator::start): Likewise.
19855 (mode_iterator::iterate_p): Likewise.
19856 (mode_iterator::get_wider): Likewise.
19857 (mode_iterator::get_known_wider): Likewise.
19858 (mode_iterator::get_2xwider): Likewise.
19859 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
19860 (FOR_EACH_MODE): Likewise.
19861 (FOR_EACH_MODE_FROM): Likewise.
19862 (FOR_EACH_MODE_UNTIL): Likewise.
19863 (FOR_EACH_WIDER_MODE): Likewise.
19864 (FOR_EACH_2XWIDER_MODE): Likewise.
19865 * builtins.c (expand_builtin_strlen): Use new mode iterators.
19866 * combine.c (simplify_comparison): Likewise
19867 * config/i386/i386.c (type_natural_mode): Likewise.
19868 * cse.c (cse_insn): Likewise.
19869 * dse.c (find_shift_sequence): Likewise.
19870 * emit-rtl.c (init_derived_machine_modes): Likewise.
19871 (init_emit_once): Likewise.
19872 * explow.c (hard_function_value): Likewise.
19873 * expmed.c (extract_fixed_bit_field_1): Likewise.
19874 (extract_bit_field_1): Likewise.
19875 (expand_divmod): Likewise.
19876 (emit_store_flag_1): Likewise.
19877 * expr.c (init_expr_target): Likewise.
19878 (convert_move): Likewise.
19879 (alignment_for_piecewise_move): Likewise.
19880 (widest_int_mode_for_size): Likewise.
19881 (emit_block_move_via_movmem): Likewise.
19882 (copy_blkmode_to_reg): Likewise.
19883 (set_storage_via_setmem): Likewise.
19884 (compress_float_constant): Likewise.
19885 * omp-low.c (omp_clause_aligned_alignment): Likewise.
19886 * optabs-query.c (get_best_extraction_insn): Likewise.
19887 * optabs.c (expand_binop): Likewise.
19888 (expand_twoval_unop): Likewise.
19889 (expand_twoval_binop): Likewise.
19890 (widen_leading): Likewise.
19891 (widen_bswap): Likewise.
19892 (expand_parity): Likewise.
19893 (expand_unop): Likewise.
19894 (prepare_cmp_insn): Likewise.
19895 (prepare_float_lib_cmp): Likewise.
19896 (expand_float): Likewise.
19897 (expand_fix): Likewise.
19898 (expand_sfix_optab): Likewise.
19899 * postreload.c (move2add_use_add2_insn): Likewise.
19900 * reg-stack.c (reg_to_stack): Likewise.
19901 * reginfo.c (choose_hard_reg_mode): Likewise.
19902 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
19903 * stor-layout.c (mode_for_size): Likewise.
19904 (smallest_mode_for_size): Likewise.
19905 (mode_for_vector): Likewise.
19906 (finish_bitfield_representative): Likewise.
19907 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
19908 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
19909 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
19910 * var-tracking.c (prepare_call_arguments): Likewise.
19912 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19913 Alan Hayward <alan.hayward@arm.com>
19914 David Sherwood <david.sherwood@arm.com>
19916 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
19917 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
19918 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
19919 * machmode.h (mode_size): Move earlier in file.
19920 (mode_precision): Likewise.
19921 (mode_inner): Likewise.
19922 (mode_nunits): Likewise.
19923 (mode_unit_size): Likewise.
19924 (unit_unit_precision): Likewise.
19925 (mode_wider): Likewise.
19926 (mode_2xwider): Likewise.
19927 (machine_mode): New class.
19928 (mode_to_bytes): New function.
19929 (mode_to_bits): Likewise.
19930 (mode_to_precision): Likewise.
19931 (mode_to_inner): Likewise.
19932 (mode_to_unit_size): Likewise.
19933 (mode_to_unit_precision): Likewise.
19934 (mode_to_nunits): Likewise.
19935 (GET_MODE_SIZE): Use mode_to_bytes.
19936 (GET_MODE_BITSIZE): Use mode_to_bits.
19937 (GET_MODE_PRECISION): Use mode_to_precision.
19938 (GET_MODE_INNER): Use mode_to_inner.
19939 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
19940 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
19941 (GET_MODE_NUNITS): Use mode_to_nunits.
19942 * system.h (ALWAYS_INLINE): New macro.
19943 * config/powerpcspe/powerpcspe-c.c
19944 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
19945 int for arg1_mode and arg2_mode.
19947 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
19948 Alan Hayward <alan.hayward@arm.com>
19949 David Sherwood <david.sherwood@arm.com>
19951 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
19952 Prefix mode names with E_ in case statements.
19953 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19954 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
19955 (aarch64_split_simd_move): Likewise.
19956 (aarch64_gen_storewb_pair): Likewise.
19957 (aarch64_gen_loadwb_pair): Likewise.
19958 (aarch64_gen_store_pair): Likewise.
19959 (aarch64_gen_load_pair): Likewise.
19960 (aarch64_get_condition_code_1): Likewise.
19961 (aarch64_constant_pool_reload_icode): Likewise.
19962 (get_rsqrte_type): Likewise.
19963 (get_rsqrts_type): Likewise.
19964 (get_recpe_type): Likewise.
19965 (get_recps_type): Likewise.
19966 (aarch64_gimplify_va_arg_expr): Likewise.
19967 (aarch64_simd_container_mode): Likewise.
19968 (aarch64_emit_load_exclusive): Likewise.
19969 (aarch64_emit_store_exclusive): Likewise.
19970 (aarch64_expand_compare_and_swap): Likewise.
19971 (aarch64_gen_atomic_cas): Likewise.
19972 (aarch64_emit_bic): Likewise.
19973 (aarch64_emit_atomic_swap): Likewise.
19974 (aarch64_emit_atomic_load_op): Likewise.
19975 (aarch64_evpc_trn): Likewise.
19976 (aarch64_evpc_uzp): Likewise.
19977 (aarch64_evpc_zip): Likewise.
19978 (aarch64_evpc_ext): Likewise.
19979 (aarch64_evpc_rev): Likewise.
19980 (aarch64_evpc_dup): Likewise.
19981 (aarch64_gen_ccmp_first): Likewise.
19982 (aarch64_gen_ccmp_next): Likewise.
19983 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
19984 (alpha_emit_xfloating_libcall): Likewise.
19985 (emit_insxl): Likewise.
19986 (alpha_arg_type): Likewise.
19987 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
19988 (arc_preferred_simd_mode): Likewise.
19989 (arc_secondary_reload): Likewise.
19990 (get_arc_condition_code): Likewise.
19991 (arc_print_operand): Likewise.
19992 (arc_legitimate_constant_p): Likewise.
19993 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19994 * config/arc/arc.md (casesi_load): Likewise.
19995 (casesi_compact_jump): Likewise.
19996 * config/arc/predicates.md (proper_comparison_operator): Likewise.
19997 (cc_use_register): Likewise.
19998 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
19999 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
20000 (arm_init_iwmmxt_builtins): Likewise.
20001 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
20002 (neon_expand_vector_init): Likewise.
20003 (arm_attr_length_move_neon): Likewise.
20004 (maybe_get_arm_condition_code): Likewise.
20005 (arm_emit_vector_const): Likewise.
20006 (arm_preferred_simd_mode): Likewise.
20007 (arm_output_iwmmxt_tinsr): Likewise.
20008 (thumb1_output_casesi): Likewise.
20009 (thumb2_output_casesi): Likewise.
20010 (arm_emit_load_exclusive): Likewise.
20011 (arm_emit_store_exclusive): Likewise.
20012 (arm_expand_compare_and_swap): Likewise.
20013 (arm_evpc_neon_vuzp): Likewise.
20014 (arm_evpc_neon_vzip): Likewise.
20015 (arm_evpc_neon_vrev): Likewise.
20016 (arm_evpc_neon_vtrn): Likewise.
20017 (arm_evpc_neon_vext): Likewise.
20018 (arm_validize_comparison): Likewise.
20019 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
20020 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
20021 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
20022 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
20023 (c6x_preferred_simd_mode): Likewise.
20024 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
20025 (epiphany_rtx_costs): Likewise.
20026 * config/epiphany/predicates.md (proper_comparison_operator):
20028 * config/frv/frv.c (condexec_memory_operand): Likewise.
20029 (frv_emit_move): Likewise.
20030 (output_move_single): Likewise.
20031 (output_condmove_single): Likewise.
20032 (frv_hard_regno_mode_ok): Likewise.
20033 (frv_matching_accg_mode): Likewise.
20034 * config/h8300/h8300.c (split_adds_subs): Likewise.
20035 (h8300_rtx_costs): Likewise.
20036 (h8300_print_operand): Likewise.
20037 (compute_mov_length): Likewise.
20038 (output_logical_op): Likewise.
20039 (compute_logical_op_length): Likewise.
20040 (compute_logical_op_cc): Likewise.
20041 (h8300_shift_needs_scratch_p): Likewise.
20042 (output_a_shift): Likewise.
20043 (compute_a_shift_length): Likewise.
20044 (compute_a_shift_cc): Likewise.
20045 (expand_a_rotate): Likewise.
20046 (output_a_rotate): Likewise.
20047 * config/i386/i386.c (classify_argument): Likewise.
20048 (function_arg_advance_32): Likewise.
20049 (function_arg_32): Likewise.
20050 (function_arg_64): Likewise.
20051 (function_value_64): Likewise.
20052 (ix86_gimplify_va_arg): Likewise.
20053 (ix86_legitimate_constant_p): Likewise.
20054 (put_condition_code): Likewise.
20055 (split_double_mode): Likewise.
20056 (ix86_avx256_split_vector_move_misalign): Likewise.
20057 (ix86_expand_vector_logical_operator): Likewise.
20058 (ix86_split_idivmod): Likewise.
20059 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
20060 (ix86_build_const_vector): Likewise.
20061 (ix86_build_signbit_mask): Likewise.
20062 (ix86_match_ccmode): Likewise.
20063 (ix86_cc_modes_compatible): Likewise.
20064 (ix86_expand_branch): Likewise.
20065 (ix86_expand_sse_cmp): Likewise.
20066 (ix86_expand_sse_movcc): Likewise.
20067 (ix86_expand_int_sse_cmp): Likewise.
20068 (ix86_expand_vec_perm_vpermi2): Likewise.
20069 (ix86_expand_vec_perm): Likewise.
20070 (ix86_expand_sse_unpack): Likewise.
20071 (ix86_expand_int_addcc): Likewise.
20072 (ix86_split_to_parts): Likewise.
20073 (ix86_vectorize_builtin_gather): Likewise.
20074 (ix86_vectorize_builtin_scatter): Likewise.
20075 (avx_vpermilp_parallel): Likewise.
20076 (inline_memory_move_cost): Likewise.
20077 (ix86_tieable_integer_mode_p): Likewise.
20078 (x86_maybe_negate_const_int): Likewise.
20079 (ix86_expand_vector_init_duplicate): Likewise.
20080 (ix86_expand_vector_init_one_nonzero): Likewise.
20081 (ix86_expand_vector_init_one_var): Likewise.
20082 (ix86_expand_vector_init_concat): Likewise.
20083 (ix86_expand_vector_init_interleave): Likewise.
20084 (ix86_expand_vector_init_general): Likewise.
20085 (ix86_expand_vector_set): Likewise.
20086 (ix86_expand_vector_extract): Likewise.
20087 (emit_reduc_half): Likewise.
20088 (ix86_emit_i387_round): Likewise.
20089 (ix86_mangle_type): Likewise.
20090 (ix86_expand_round_sse4): Likewise.
20091 (expand_vec_perm_blend): Likewise.
20092 (canonicalize_vector_int_perm): Likewise.
20093 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
20094 (expand_vec_perm_1): Likewise.
20095 (expand_vec_perm_interleave3): Likewise.
20096 (expand_vec_perm_even_odd_pack): Likewise.
20097 (expand_vec_perm_even_odd_1): Likewise.
20098 (expand_vec_perm_broadcast_1): Likewise.
20099 (ix86_vectorize_vec_perm_const_ok): Likewise.
20100 (ix86_expand_vecop_qihi): Likewise.
20101 (ix86_expand_mul_widen_hilo): Likewise.
20102 (ix86_expand_sse2_abs): Likewise.
20103 (ix86_expand_pextr): Likewise.
20104 (ix86_expand_pinsr): Likewise.
20105 (ix86_preferred_simd_mode): Likewise.
20106 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
20107 * config/i386/sse.md (*andnot<mode>3): Likewise.
20108 (<mask_codefor><code><mode>3<mask_name>): Likewise.
20109 (*<code><mode>3): Likewise.
20110 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
20111 (ia64_expand_atomic_op): Likewise.
20112 (ia64_arg_type): Likewise.
20113 (ia64_mode_to_int): Likewise.
20114 (ia64_scalar_mode_supported_p): Likewise.
20115 (ia64_vector_mode_supported_p): Likewise.
20116 (expand_vec_perm_broadcast): Likewise.
20117 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
20118 (iq2000_function_arg_advance): Likewise.
20119 (iq2000_function_arg): Likewise.
20120 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
20121 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
20122 (m68k_libcall_value): Likewise.
20123 (m68k_function_value): Likewise.
20124 (sched_attr_op_type): Likewise.
20125 * config/mcore/mcore.c (mcore_output_move): Likewise.
20126 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
20128 (microblaze_function_arg): Likewise.
20129 * config/mips/mips.c (mips16_build_call_stub): Likewise.
20130 (mips_print_operand): Likewise.
20131 (mips_mode_ok_for_mov_fmt_p): Likewise.
20132 (mips_vector_mode_supported_p): Likewise.
20133 (mips_preferred_simd_mode): Likewise.
20134 (mips_expand_vpc_loongson_even_odd): Likewise.
20135 (mips_expand_vec_unpack): Likewise.
20136 (mips_expand_vi_broadcast): Likewise.
20137 (mips_expand_vector_init): Likewise.
20138 (mips_expand_vec_reduc): Likewise.
20139 (mips_expand_msa_cmp): Likewise.
20140 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
20141 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
20142 (cc_flags_for_mode): Likewise.
20143 * config/msp430/msp430.c (msp430_print_operand): Likewise.
20144 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
20145 (nds32_output_casesi_pc_relative): Likewise.
20146 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20147 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
20148 (nvptx_gen_unpack): Likewise.
20149 (nvptx_gen_pack): Likewise.
20150 (nvptx_gen_shuffle): Likewise.
20151 (nvptx_gen_wcast): Likewise.
20152 (nvptx_preferred_simd_mode): Likewise.
20153 * config/pa/pa.c (pa_secondary_reload): Likewise.
20154 * config/pa/predicates.md (base14_operand): Likewise.
20155 * config/powerpcspe/powerpcspe-c.c
20156 (altivec_resolve_overloaded_builtin): Likewise.
20157 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
20159 (rs6000_preferred_simd_mode): Likewise.
20160 (output_vec_const_move): Likewise.
20161 (rs6000_expand_vector_extract): Likewise.
20162 (rs6000_split_vec_extract_var): Likewise.
20163 (reg_offset_addressing_ok_p): Likewise.
20164 (rs6000_legitimate_offset_address_p): Likewise.
20165 (rs6000_legitimize_address): Likewise.
20166 (rs6000_emit_set_const): Likewise.
20167 (rs6000_const_vec): Likewise.
20168 (rs6000_emit_move): Likewise.
20169 (spe_build_register_parallel): Likewise.
20170 (rs6000_darwin64_record_arg_recurse): Likewise.
20171 (swap_selector_for_mode): Likewise.
20172 (spe_init_builtins): Likewise.
20173 (paired_init_builtins): Likewise.
20174 (altivec_init_builtins): Likewise.
20175 (do_load_for_compare): Likewise.
20176 (rs6000_generate_compare): Likewise.
20177 (rs6000_expand_float128_convert): Likewise.
20178 (emit_load_locked): Likewise.
20179 (emit_store_conditional): Likewise.
20180 (rs6000_output_function_epilogue): Likewise.
20181 (rs6000_handle_altivec_attribute): Likewise.
20182 (rs6000_function_value): Likewise.
20183 (emit_fusion_gpr_load): Likewise.
20184 (emit_fusion_p9_load): Likewise.
20185 (emit_fusion_p9_store): Likewise.
20186 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
20187 (fusion_gpr_mem_load): Likewise.
20188 (fusion_addis_mem_combo_load): Likewise.
20189 (fusion_addis_mem_combo_store): Likewise.
20190 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
20191 (fusion_gpr_mem_load): Likewise.
20192 (fusion_addis_mem_combo_load): Likewise.
20193 (fusion_addis_mem_combo_store): Likewise.
20194 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20196 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
20197 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
20198 (rs6000_preferred_simd_mode): Likewise.
20199 (output_vec_const_move): Likewise.
20200 (rs6000_expand_vector_extract): Likewise.
20201 (rs6000_split_vec_extract_var): Likewise.
20202 (reg_offset_addressing_ok_p): Likewise.
20203 (rs6000_legitimate_offset_address_p): Likewise.
20204 (rs6000_legitimize_address): Likewise.
20205 (rs6000_emit_set_const): Likewise.
20206 (rs6000_const_vec): Likewise.
20207 (rs6000_emit_move): Likewise.
20208 (rs6000_darwin64_record_arg_recurse): Likewise.
20209 (swap_selector_for_mode): Likewise.
20210 (paired_init_builtins): Likewise.
20211 (altivec_init_builtins): Likewise.
20212 (rs6000_expand_float128_convert): Likewise.
20213 (emit_load_locked): Likewise.
20214 (emit_store_conditional): Likewise.
20215 (rs6000_output_function_epilogue): Likewise.
20216 (rs6000_handle_altivec_attribute): Likewise.
20217 (rs6000_function_value): Likewise.
20218 (emit_fusion_gpr_load): Likewise.
20219 (emit_fusion_p9_load): Likewise.
20220 (emit_fusion_p9_store): Likewise.
20221 * config/rx/rx.c (rx_gen_move_template): Likewise.
20222 (flags_from_mode): Likewise.
20223 * config/s390/predicates.md (s390_alc_comparison): Likewise.
20224 (s390_slb_comparison): Likewise.
20225 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
20226 (s390_vector_mode_supported_p): Likewise.
20227 (s390_cc_modes_compatible): Likewise.
20228 (s390_match_ccmode_set): Likewise.
20229 (s390_canonicalize_comparison): Likewise.
20230 (s390_emit_compare_and_swap): Likewise.
20231 (s390_branch_condition_mask): Likewise.
20232 (s390_rtx_costs): Likewise.
20233 (s390_secondary_reload): Likewise.
20234 (__SECONDARY_RELOAD_CASE): Likewise.
20235 (s390_expand_cs): Likewise.
20236 (s390_preferred_simd_mode): Likewise.
20237 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
20238 * config/sh/sh.c (sh_print_operand): Likewise.
20239 (dump_table): Likewise.
20240 (sh_secondary_reload): Likewise.
20241 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20242 * config/sh/sh.md (casesi_worker_1): Likewise.
20243 (casesi_worker_2): Likewise.
20244 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
20245 (fcc_comparison_operator): Likewise.
20246 * config/sparc/sparc.c (sparc_expand_move): Likewise.
20247 (emit_soft_tfmode_cvt): Likewise.
20248 (sparc_preferred_simd_mode): Likewise.
20249 (output_cbranch): Likewise.
20250 (sparc_print_operand): Likewise.
20251 (sparc_expand_vec_perm_bmask): Likewise.
20252 (vector_init_bshuffle): Likewise.
20253 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
20254 (spu_vector_mode_supported_p): Likewise.
20255 (spu_expand_insv): Likewise.
20256 (spu_emit_branch_or_set): Likewise.
20257 (spu_handle_vector_attribute): Likewise.
20258 (spu_builtin_splats): Likewise.
20259 (spu_builtin_extract): Likewise.
20260 (spu_builtin_promote): Likewise.
20261 (spu_expand_sign_extend): Likewise.
20262 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
20263 (tilegx_simd_int): Likewise.
20264 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
20265 (tilepro_simd_int): Likewise.
20266 * config/v850/v850.c (const_double_split): Likewise.
20267 (v850_print_operand): Likewise.
20268 (ep_memory_offset): Likewise.
20269 * config/vax/vax.c (vax_rtx_costs): Likewise.
20270 (vax_output_int_move): Likewise.
20271 (vax_output_int_add): Likewise.
20272 (vax_output_int_subtract): Likewise.
20273 * config/visium/predicates.md (visium_branch_operator): Likewise.
20274 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
20275 (visium_print_operand_address): Likewise.
20276 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
20277 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
20278 (xtensa_expand_conditional_branch): Likewise.
20279 (xtensa_copy_incoming_a7): Likewise.
20280 (xtensa_output_literal): Likewise.
20281 * dfp.c (decimal_real_maxval): Likewise.
20282 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
20284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20285 Alan Hayward <alan.hayward@arm.com>
20286 David Sherwood <david.sherwood@arm.com>
20288 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
20289 (mode_nunits_inline): Likewise.
20290 (mode_inner_inline): Likewise.
20291 (mode_unit_size_inline): Likewise.
20292 (mode_unit_precision_inline): Likewise.
20293 (emit_insn_modes_h): Likewise. Also emit a #define of the
20295 (emit_mode_wider): Add an E_ prefix to mode names.
20296 (emit_mode_complex): Likewise.
20297 (emit_mode_inner): Likewise.
20298 (emit_mode_adjustments): Likewise.
20299 (emit_mode_int_n): Likewise.
20300 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
20301 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
20302 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
20303 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
20304 (CRC32_BUILTIN, ENTRY): Likewise.
20305 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
20306 (aarch64_pop_regs): Likewise.
20307 (aarch64_process_components): Likewise.
20308 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
20309 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
20310 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
20311 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
20312 * config/arm/arm.c (arm_init_libfuncs): Likewise.
20313 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
20315 * config/i386/i386-builtin.def (pcmpestr): Likewise.
20316 (pcmpistr): Likewise.
20317 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
20318 * config/mmix/mmix.c (mmix_output_condition): Likewise.
20319 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
20321 * config/rl78/rl78.c (mduc_regs): Likewise.
20322 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20323 (htm_expand_builtin): Likewise.
20324 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
20325 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
20326 * config/xtensa/xtensa.c (print_operand): Likewise.
20327 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
20328 (NUM_MODE_VECTOR_INT): Likewise.
20329 * genoutput.c (null_operand): Likewise.
20330 (output_operand_data): Likewise.
20331 * genrecog.c (print_parameter_value): Likewise.
20332 * lra.c (debug_operand_data): Likewise.
20334 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20336 * dwarf2out.c (scompare_loc_descriptor_wide)
20337 (scompare_loc_descriptor_narrow): New functions, split out from...
20338 (scompare_loc_descriptor): ...here.
20339 * expmed.c (emit_store_flag_int): New function, split out from...
20340 (emit_store_flag): ...here.
20342 2017-08-30 Richard Biener <rguenther@suse.de>
20344 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
20345 (dwarf2out_early_finish): Move setting of AT_pubnames from
20346 early debug output to early finish.
20348 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20350 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
20351 and -mdata-region to the assembler.
20353 2017-08-30 Richard Biener <rguenther@suse.de>
20355 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
20357 (gen_subprogram_die): Add DW_AT_object_pointer only early.
20358 (dwarf2out_early_global_decl): Only generate a DIE for the
20359 abstract origin if it doesn't already exist or is a declaration DIE.
20360 (resolve_addr): Do not add the linkage name twice when
20361 generating a stub DIE for the DW_TAG_GNU_call_site target.
20363 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
20365 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20366 Use machine_mode rather than int for arg1_mode.
20368 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
20371 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
20372 that the second argument of the built-in functions to unpack
20373 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
20374 switch statement instead a lot of if statements.
20375 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
20376 Allow 64-bit values to be in Altivec registers as well as
20377 traditional floating point registers.
20378 (pack<mode>, FMOVE128_VSX iterator): Likewise.
20380 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
20382 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
20383 MAX_REGS_PER_ADDRESS == 1.
20385 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
20387 * config/i386/i386.opt (flag_fentry): Do not init to -1.
20388 * config/i386/i386.c (ix86_option_override_internal): Simplify
20389 setting of opts->x_flag_entry.
20391 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20392 Jakub Jelinek <jakub@redhat.com>
20393 Richard Biener <rguenther@suse.de>
20395 PR tree-optimization/81503
20396 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
20397 folded constant fits in the target type; reorder tests for clarity.
20399 2017-08-29 Martin Liska <mliska@suse.cz>
20401 * passes.def: Include pass_lower_switch.
20402 * stmt.c (dump_case_nodes): Remove and move to
20403 tree-switch-conversion.
20404 (case_values_threshold): Likewise.
20405 (expand_switch_as_decision_tree_p): Likewise.
20406 (emit_case_decision_tree): Likewise.
20407 (expand_case): Likewise.
20408 (balance_case_nodes): Likewise.
20409 (node_has_low_bound): Likewise.
20410 (node_has_high_bound): Likewise.
20411 (node_is_bounded): Likewise.
20412 (emit_case_nodes): Likewise.
20413 (struct simple_case_node): New struct.
20414 (add_case_node): Remove.
20415 (emit_case_dispatch_table): Use vector instead of case_list.
20416 (reset_out_edges_aux): Remove.
20417 (compute_cases_per_edge): Likewise.
20418 (expand_case): Build list of simple_case_node.
20419 (expand_sjlj_dispatch_table): Use it.
20420 * tree-switch-conversion.c (struct case_node): Moved from
20421 stmt.c and adjusted.
20422 (emit_case_nodes): Likewise.
20423 (node_has_low_bound): Likewise.
20424 (node_has_high_bound): Likewise.
20425 (node_is_bounded): Likewise.
20426 (case_values_threshold): Likewise.
20427 (reset_out_edges_aux): Likewise.
20428 (compute_cases_per_edge): Likewise.
20429 (add_case_node): Likewise.
20430 (dump_case_nodes): Likewise.
20431 (balance_case_nodes): Likewise.
20432 (expand_switch_as_decision_tree_p): Likewise.
20433 (emit_jump): Likewise.
20434 (emit_case_decision_tree): Likewise.
20435 (try_switch_expansion): Likewise.
20436 (do_jump_if_equal): Likewise.
20437 (emit_cmp_and_jump_insns): Likewise.
20438 (fix_phi_operands_for_edge): New function.
20439 (record_phi_operand_mapping): Likewise.
20440 (class pass_lower_switch): New pass.
20441 (pass_lower_switch::execute): New function.
20442 (make_pass_lower_switch): Likewise.
20443 (conditional_probability):
20444 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
20445 * tree-pass.h: Add make_pass_lower_switch.
20447 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20450 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
20453 2017-08-29 Richard Biener <rguenther@suse.de>
20455 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
20456 we do not add a DW_AT_inline attribute twice.
20457 (gen_subprogram_die): Remove code setting DW_AT_inline on
20458 DECL_ABSTRACT_P nodes.
20460 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
20462 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
20463 calls to internal functions.
20464 (gimplify_modify_expr): Likewise.
20465 * tree-call-cdce.c (use_internal_fn): Likewise.
20466 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
20467 (convert_to_divmod): Set the nothrow flag.
20468 * tree-if-conv.c (predicate_mem_writes): Likewise.
20469 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20470 (vectorizable_call): Likewise.
20471 (vectorizable_store): Likewise.
20472 (vectorizable_load): Likewise.
20473 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
20474 (vect_recog_mask_conversion_pattern): Likewise.
20476 2017-08-29 Martin Liska <mliska@suse.cz>
20479 * gcc.c (driver_handle_option): Add new argument.
20480 * opts-common.c (handle_option): Pass
20481 target_option_override_hook.
20482 * opts-global.c (lang_handle_option): Add new option.
20483 (set_default_handlers): Add new argument.
20484 (decode_options): Likewise.
20485 * opts.c (target_handle_option): Likewise.
20486 (common_handle_option): Call target_option_override_hook.
20487 * opts.h (struct cl_option_handler_func): Add hook for
20488 target option override.
20489 (struct cl_option_handlers): Likewise.
20490 (set_default_handlers): Add new argument.
20491 (decode_options): Likewise.
20492 (common_handle_option): Likewise.
20493 (target_handle_option): Likewise.
20494 * toplev.c (toplev::main): Pass targetm.target_option.override
20497 2017-08-29 Richard Biener <rguenther@suse.de>
20498 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
20500 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
20501 life to the active subtree.
20503 2017-08-28 Jeff Law <law@redhat.com>
20505 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
20506 derive_equivalences.
20507 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
20508 Code moved into....
20509 (edge_info::derive_equivalences): New private member function
20511 * tree-ssa-dom.c (class edge_info): Changed from a struct
20512 to a class. Add ctor/dtor, methods and data members.
20513 (edge_info::edge_info): Renamed from allocate_edge_info.
20514 Initialize additional members.
20515 (edge_info::~edge_info): New.
20516 (free_dom_edge_info): Delete the edge info.
20517 (record_edge_info): Use new class & associated member functions.
20518 Tighten forms for testing for edge equivalences.
20519 (record_temporary_equivalences): Iterate over the simple
20520 equivalences rather than assuming there's only one per edge.
20521 (cprop_into_successor_phis): Iterate over the simple
20522 equivalences rather than assuming there's only one per edge.
20523 (optimize_stmt): Use operand_equal_p rather than pointer
20524 equality for mini-DSE code.
20526 2017-08-28 Nathan Sidwell <nathan@acm.org>
20528 * gcc.c (execute): Fold SIGPIPE handling into switch
20529 statement. Adjust internal error message.
20531 2017-08-28 Richard Biener <rguenther@suse.de>
20534 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
20535 Do nothing for removed DIEs.
20537 2017-08-28 Richard Biener <rguenther@suse.de>
20539 PR tree-optimization/81977
20540 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
20543 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
20546 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
20547 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
20548 using targetm.gen_mem_thread_fence.
20550 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
20553 * config/i386/i386.md (*<btsc><mode>): Change operand 2
20554 predicate to register_operand. Reorder operands.
20555 (*btr<mode>): Ditto.
20556 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
20557 (*btr<mode>_mask): Ditto.
20559 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
20561 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
20562 * config/rs6000/xmmintrin.h: New file.
20563 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
20565 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20568 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
20569 parameter and_insn and return it.
20570 (recombine_lvx_pattern): Insert a copy to ensure availability of
20571 the base register of the copied masking operation at the point of
20572 the instruction replacement.
20573 (recombine_stvx_pattern): Likewise.
20575 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20577 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
20578 undocumented switches.
20579 (-mpower9-dform-vector): Likewise.
20580 (-mpower9-dform): Likewise.
20581 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
20582 comments to delete references to -mpower9-dform* switches.
20583 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
20584 Delete reference to -mpower9-dform* switches, test for
20585 -mpower9-vector instead.
20586 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
20587 (OTHER_P9_VECTOR_MASKS): Likewise.
20588 (POWERPC_MASKS): Likewise.
20589 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
20590 tests against -mpower9-dform* to -mpower9-vector. Delete code
20591 that checked for -mpower9-dform* consistancy with other options.
20592 Add test for -mpower9-misc to enable other power9 switches.
20593 (rs6000_init_hard_regno_mode_ok): Likewise.
20594 (rs6000_option_override_internal): Likewise.
20595 (rs6000_emit_prologue): Likewise.
20596 (rs6000_emit_epilogue): Likewise.
20597 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
20598 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
20599 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
20601 (emit_fusion_p9_store): Likewise.
20602 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
20603 resetting these macros if the assembler does not support ISA 3.0
20605 (TARGET_P9_DFORM_VECTOR): Likewise.
20606 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
20607 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
20609 2017-08-25 Alan Modra <amodra@gmail.com>
20611 PR rtl-optimization/81747
20612 * cse.c (cse_extended_basic_block): Don't attempt to record
20613 equivalences for degenerate conditional branches that branch
20614 to their fall-through.
20616 2017-08-24 Martin Sebor <msebor@redhat.com>
20618 PR middle-end/81908
20619 * gimple-fold.c (size_must_be_zero_p): New function.
20620 (gimple_fold_builtin_memory_op): Call it.
20622 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
20624 * config/rs6000/mm_malloc.h: New file.
20626 2017-08-24 Bin Cheng <bin.cheng@arm.com>
20628 PR tree-optimization/81913
20629 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
20630 analysis when either IVs in condition can wrap.
20632 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
20634 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
20635 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
20637 2017-08-24 Richard Biener <rguenther@suse.de>
20640 * targhooks.c (default_target_can_inline_p): Properly
20641 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
20642 is present and always compare.
20643 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
20644 infer -mfpmath=sse from TARGET_SSE_P.
20645 (ix86_can_inline_p): Properly use target_option_default_node when
20646 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
20648 2017-08-24 Richard Biener <rguenther@suse.de>
20651 * dwarf2out.c (output_die): Handle flag_generate_offload like
20653 (output_comp_unit): Likewise.
20654 (gen_array_type_die): Likewise.
20655 (dwarf2out_early_finish): Likewise.
20656 (note_variable_value_in_expr): Likewise.
20657 (dwarf2out_finish): Likewise. Adjust assert.
20658 * cgraphunit.c (symbol_table::compile): Move setting of
20659 flag_generate_offload earlier ...
20660 (symbol_table::finalize_compilation_unit): ... here, before
20661 early debug finalization.
20663 2017-08-24 Richard Biener <rguenther@suse.de>
20665 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
20666 and ipa-fnsummary.h.
20667 (ix86_can_inline_p): When ix86_fpmath flags do not match
20668 check whether the callee uses FP math at all.
20670 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
20672 PR middle-end/81931
20673 * tree-ssanames.c (get_nonzero_bits): Use element_precision
20674 instead of TYPE_PRECISION.
20676 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
20677 Alan Hayward <alan.hayward@arm.com>
20678 David Sherwood <david.sherwood@arm.com>
20680 * combine.c (make_extraction): Use subreg_offset_from_lsb.
20682 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
20684 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
20686 (ix86_frame::stack_realign_allocate): New field.
20687 (struct machine_frame_state): Modify comments.
20688 (machine_frame_state::sp_realigned_fp_end): New field.
20689 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
20690 layout calculation.
20691 (sp_valid_at): Add assertion to assure no attempt to access invalid
20692 offset of a realigned stack.
20693 (fp_valid_at): Likewise.
20694 (choose_baseaddr): Modify comments.
20695 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
20696 ix86_expand_prologue.
20697 (ix86_expand_prologue): Modify stack realignment and allocation.
20698 (ix86_expand_epilogue): Modify comments.
20699 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
20700 avx2_runtime, avx512f, and avx512f_runtime.
20702 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
20704 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
20705 (mstackrealign): Do not init to -1.
20706 * config/i386/i386.c (ix86_option_override_internal):
20707 Check opts_set, not opts when setting default value of
20708 opts->x_ix86_force_align_arg_pointer.
20710 2017-08-23 Richard Biener <rguenther@suse.de>
20712 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
20713 lang_hooks.decl_printable_name.
20714 * print-rtl-function.c (print_rtx_function): Likewise.
20715 * tree-pretty-print.c (dump_function_header): Likewise.
20717 2017-08-23 Richard Biener <rguenther@suse.de>
20720 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
20721 -g0 at compile-time.
20723 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20725 PR middle-end/19706
20726 * doc/sourcebuild.texi (Other hardware attributes):
20729 2017-08-23 Tamar Christina <tamar.christina@arm.com>
20731 PR middle-end/19706
20732 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
20733 Fix single-use check.
20735 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
20737 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
20739 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20741 * doc/install.texi: Modify to add more details on running selected
20744 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
20746 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
20747 is combined with -mabi=ms.
20748 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
20751 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20753 PR tree-optimization/81488
20754 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
20755 and cached_basis fields.
20756 (MAX_SPREAD): New constant.
20757 (alloc_cand_and_find_basis): Initialize new fields.
20758 (clear_visited): New function.
20759 (create_phi_basis_1): Rename from create_phi_basis, set visited
20760 and cached_basis fields.
20761 (create_phi_basis): New wrapper function.
20762 (phi_add_costs_1): Rename from phi_add_costs, add spread
20763 parameter, set visited field, short-circuit when limits reached.
20764 (phi_add_costs): New wrapper function.
20765 (record_phi_increments_1): Rename from record_phi_increments, set
20767 (record_phi_increments): New wrapper function.
20768 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
20769 (phi_incr_cost): New wrapper function.
20770 (all_phi_incrs_profitable_1): Rename from
20771 all_phi_incrs_profitable, set visited field.
20772 (all_phi_incrs_profitable): New wrapper function.
20774 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20775 Alan Hayward <alan.hayward@arm.com>
20776 David Sherwood <david.sherwood@arm.com>
20778 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
20779 that takes the outer and inner modes.
20780 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
20781 comparison as the canonical test for a paradoxical subreg.
20782 * combine.c (simplify_set): Use paradoxical_subreg_p.
20783 (make_extraction): Likewise.
20784 (force_to_mode): Likewise.
20785 (rtx_equal_for_field_assignment_p): Likewise.
20786 (gen_lowpart_for_combine): Likewise.
20787 (simplify_comparison): Likewise.
20788 * cse.c (equiv_constant): Likewise.
20789 * expmed.c (store_bit_field_1): Likewise.
20790 * final.c (alter_subreg): Likewise.
20791 * fwprop.c (propagate_rtx): Likewise.
20792 (forward_propagate_subreg): Likewise.
20793 * ira-conflicts.c (ira_build_conflicts): Likewise.
20794 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
20795 * lra-constraints.c (curr_insn_transform): Likewise.
20796 (split_reg): Likewise.
20797 * lra-eliminations.c (move_plus_up): Likewise.
20798 (lra_eliminate_regs_1): Likewise.
20799 * recog.c (general_operand): Likewise.
20800 * ree.c (combine_reaching_defs): Likewise.
20801 * reload.c (push_reload): Likewise.
20802 (find_reloads): Likewise.
20803 * reload1.c (elimination_effects): Likewise.
20804 (compute_reload_subreg_offset): Likewise.
20805 (choose_reload_regs): Likewise.
20806 * rtlanal.c (subreg_lsb_1): Likewise.
20807 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
20808 (simplify_subreg): Likewise.
20809 * var-tracking.c (track_loc_p): Likewise.
20810 * emit-rtl.c (byte_lowpart_offset): Likewise.
20811 (paradoxical_subreg_p): Delete out-of-line definition.
20813 2017-08-22 Jeff Law <law@redhat.com>
20815 PR tree-optimization/81741
20816 PR tree-optimization/71947
20817 * tree-ssa-dom.c: Include tree-inline.h.
20818 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
20819 equivalences if one is more expensive to compute than the other.
20820 * tree-ssa-scopedtables.h (class const_or_copies): Make
20821 record_const_or_copy_raw method private.
20822 (class avail_exprs_stack): New method simplify_binary_operation.
20823 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
20824 avail_exprs_stack::simplify_binary_operation as needed.
20825 (avail_exprs_stack::simplify_binary_operation): New function.
20827 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
20829 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
20830 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
20831 (DOT_SYMBOLS): Likewise.
20832 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
20833 (RELOCATABLE_NEEDS_FIXUP): Likewise.
20834 (RS6000_ABI_NAME): Likewise.
20835 (TARGET_CMODEL): Likewise.
20836 (TOC_SECTION_ASM_OP): Likewise.
20837 (SET_CMODEL): New macro.
20838 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
20840 2017-08-22 Richard Biener <rguenther@suse.de>
20842 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
20843 to allow for free-lang-data replacements similar to verify_type_variant.
20845 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
20846 Alan Hayward <alan.hayward@arm.com>
20847 David Sherwood <david.sherwood@arm.com>
20849 * config/aarch64/aarch64.md (casesi): Use DImode rather than
20850 VOIDmode for the LABEL_REF.
20852 2017-08-22 Richard Biener <rguenther@suse.de>
20854 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
20856 2017-08-22 Richard Biener <rguenther@suse.de>
20858 * common.opt (feliminate-dwarf2-dups): Ignore.
20859 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
20860 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
20861 same_die_p_wrap, compute_section_prefix,
20862 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
20863 (comdat_symbol_id, comdat_symbol_number): Likewise.
20864 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
20866 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
20867 (output_die): Mark unreachable path unreachable.
20868 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
20869 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
20870 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
20871 (dwarf2out_early_finish): Likewise.
20873 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
20875 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
20877 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
20880 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
20881 not VAR_P. Filter attribute warnings with OPT_Wattributes.
20882 (avr_attribute_table) <io, io_low, address>: Initialize
20883 .decl_required with true.
20885 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
20887 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
20888 undocumented debugging options.
20889 (-mvsx-scalar-double): Likewise.
20890 (-mallow-df-permute): Likewise.
20891 (-mvectorize-builtins): Likewise.
20892 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
20893 (rs6000_builtin_vectorized_function): Likewise.
20894 (rs6000_builtin_md_vectorized_function): Likewise.
20895 (rs6000_opt_vars): Likewise.
20897 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
20900 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
20901 (*btrq_imm): Rename from *btrq.
20902 (*btcq_imm): Rename from *btcq.
20903 (btsc): New code attribute.
20904 (*<btsc><mode>): New insn pattern.
20905 (*btr<mode>): Ditto.
20906 (*<btsc><mode>_mask): New insn_and_split pattern.
20907 (*btr<mode>_mask): Ditto.
20909 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20910 Alan Hayward <alan.hayward@arm.com>
20911 David Sherwood <david.sherwood@arm.com>
20913 * function.c (pad_below): Simplify padding calculation.
20915 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20916 Alan Hayward <alan.hayward@arm.com>
20917 David Sherwood <david.sherwood@arm.com>
20919 * target.def (function_prologue): Remove frame size argument.
20920 (function_epilogue): Likewise.
20921 * doc/tm.texi: Regenerate.
20922 * final.c (final_start_function): Update call to function_prologue.
20923 (final_end_function): Update call to function_epilogue.
20924 (default_function_pro_epilogue): Remove frame size argument.
20925 * output.h (default_function_pro_epilogue): Likewise.
20926 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
20927 (arm_output_function_prologue): Likewise.
20928 * config/frv/frv.c (frv_function_prologue): Likewise.
20929 (frv_function_epilogue): Likewise.
20930 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
20931 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
20932 (ia64_output_function_epilogue): Likewise.
20933 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
20934 (m32r_output_function_epilogue): Likewise.
20935 * config/microblaze/microblaze.c (microblaze_function_prologue)
20936 (microblaze_function_epilogue): Likewise.
20937 * config/mips/mips.c (mips_output_function_prologue): Likewise.
20938 (mips_output_function_epilogue): Likewise.
20939 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
20940 (mmix_target_asm_function_epilogue): Likewise.
20941 * config/msp430/msp430.c (msp430_start_function): Likewise.
20942 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
20943 (nds32_asm_function_epilogue): Likewise.
20944 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
20945 * config/pa/pa.c (pa_output_function_prologue): Likewise.
20946 (pa_output_function_epilogue): Likewise.
20947 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
20948 (rs6000_output_function_epilogue): Likewise.
20949 * config/rl78/rl78.c (rl78_start_function): Likewise.
20950 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
20951 (rs6000_output_function_epilogue): Likewise.
20952 * config/rx/rx.c (rx_output_function_prologue): Likewise.
20953 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
20954 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
20955 (sparc_asm_function_epilogue): Likewise.
20957 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20959 * tree.h (type_has_mode_precision_p): New function.
20960 * convert.c (convert_to_integer_1): Use it.
20961 * expr.c (expand_expr_real_2): Likewise.
20962 (expand_expr_real_1): Likewise.
20963 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
20964 * match.pd: Likewise.
20965 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
20966 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
20967 * tree-tailcall.c (process_assignment): Likewise.
20968 * tree-vect-loop.c (vectorizable_reduction): Likewise.
20969 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
20970 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
20971 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
20972 (vectorizable_assignment): Likewise.
20973 (vectorizable_shift): Likewise.
20974 (vectorizable_operation): Likewise.
20975 * tree-vrp.c (register_edge_assert_for_2): Likewise.
20977 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
20979 * match.pd: Add pow (C, x) simplification.
20981 2017-08-21 Richard Biener <rguenther@suse.de>
20983 PR tree-optimization/81900
20984 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
20985 for blocks with abnormal predecessors.
20986 (compute_antic): Do not set visited flag prematurely.
20988 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
20991 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
20993 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
20995 * stor-layout.h (vector_type_mode): Move to...
20996 * tree.h (vector_type_mode): ...here.
20997 * stor-layout.c (vector_type_mode): Move to...
20998 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
21000 2017-08-21 Richard Biener <rguenther@suse.de>
21002 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
21003 register_external_die hooks.
21004 (debug_false_tree_charstarstar_uhwistar): Declare.
21005 (debug_nothing_tree_charstar_uhwi): Likewise.
21006 * debug.c (do_nothing_debug_hooks): Adjust.
21007 (debug_false_tree_charstarstar_uhwistar): New do nothing.
21008 (debug_nothing_tree_charstar_uhwi): Likewise.
21009 * dbxout.c (dbx_debug_hooks): Adjust.
21010 (xcoff_debug_hooks): Likewise.
21011 * sdbout.c (sdb_debug_hooks): Likewise.
21012 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
21013 * dwarf2out.c (macinfo_label_base): New global.
21014 (dwarf2out_register_external_die): New function for the
21015 register_external_die hook.
21016 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
21017 (dwarf2_debug_hooks): Use them.
21018 (dwarf2_lineno_debug_hooks): Adjust.
21019 (struct die_struct): Add with_offset flag.
21020 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
21021 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
21022 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
21023 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
21024 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
21025 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
21026 defining section names for the early LTO debug variants.
21027 (reset_indirect_string): New helper.
21028 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
21029 (print_dw_val): Add support for offsetted symbol references.
21030 (get_ultimate_context): Split out from is_cxx.
21031 (is_cxx): Use get_ultimate_context.
21032 (is_fortran): Add decl overload.
21033 (compute_comp_unit_symbol): Split out worker from
21034 compute_section_prefix.
21035 (compute_section_prefix): Call compute_comp_unit_symbol and
21036 set comdat_type_p here.
21037 (output_die): Skip DIE symbol output for the LTO added one.
21038 Handle DIE symbol references with offset.
21039 (output_comp_unit): Guard section name mangling properly.
21040 For LTO debug sections emit a symbol at the section beginning
21041 which we use to refer to its DIEs.
21042 (add_abstract_origin_attribute): For DIEs registered via
21043 dwarf2out_register_external_die directly refer to the early
21044 DIE rather than indirectly through the shadow one we created.
21045 Remove obsolete call to dwarf2out_abstract_function for
21046 non-function/block origins.
21047 (gen_array_type_die): When generating early LTO debug do
21048 not emit DW_AT_string_length.
21049 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
21050 late when in LTO. As suggested place a gcc_unreachable for
21051 the DECL_ABSTRACT_P case.
21052 (gen_subprogram_die): Avoid another specification DIE
21053 for early built declarations/definitions for the late LTO case.
21054 (gen_variable_die): Add type references for late duplicated VLA dies
21056 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
21057 we have the abstract instance already.
21058 (process_scope_var): Adjust decl DIE contexts in LTO which
21059 first puts them in limbo.
21060 (gen_decl_die): Do not generate type DIEs late apart from
21061 types for VLAs or for decls we do not yet have a DIE. Do not
21062 call dwarf2out_abstract_function late.
21063 (dwarf2out_early_global_decl): Make sure to create DIEs
21064 for abstract instances of a decl first.
21065 (dwarf2out_late_global_decl): Adjust comment.
21066 (output_macinfo_op): With multiple macro sections use
21067 macinfo_label_base to distinguish labels.
21068 (output_macinfo): Likewise. Update macinfo_label_base.
21069 Pass in the line info label.
21070 (note_variable_value_in_expr): When generating LTO resolve
21071 all variable values here by generating DIEs as needed.
21072 (init_sections_and_labels): Add early LTO debug flag parameter
21073 and generate different sections and names if set. Add generation
21074 counter for the labels so we can have multiple of them.
21075 (reset_dies): Helper to allow DIEs to be output multiple times.
21076 (dwarf2out_finish): When outputting DIEs to the fat part of an
21077 LTO object first reset DIEs.
21078 (dwarf2out_early_finish): Output early DIEs when generating LTO.
21079 (modified_type_die): Check for decl_ultimate_origin being self
21081 (gen_type_die_with_usage): Likewise.
21082 (gen_typedef_die): Allow decl_ultimate_origin being self.
21083 (set_decl_abstract_flags): Remove.
21084 (set_block_abstract_flags): Likewise.
21085 (dwarf2out_abstract_function): Treat the early generated DIEs
21086 as the abstract copy and only add DW_AT_inline and
21087 DW_AT_artificial here and call set_decl_origin_self.
21088 If the DIE has an abstract origin don't do anything.
21089 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
21090 if we have none yet (Go fails to build one, PR78628).
21091 (variably_modified_type_p): Prevent endless recursion for Ada
21092 cyclic pointer types.
21093 * lto-streamer-in.c: Include debug.h.
21094 (dref_queue): New global.
21095 (lto_read_tree_1): Stream in DIE references.
21096 (lto_input_tree): Register DIE references.
21097 (input_function): Stream DECL_DEBUG_ARGS.
21098 * lto-streamer-out.c: Include debug.h.
21099 (lto_write_tree_1): Output DIE references.
21100 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
21101 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
21102 (output_function): Stream DECL_DEBUG_ARGS.
21103 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
21104 Stream DECL_ABSTRACT_ORIGIN.
21105 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
21106 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
21107 DECL_CONTEXT for file-scope decls.
21108 * lto-streamer.h (struct dref_entry): Declare.
21109 (dref_queue): Likewise.
21110 * cfgexpand.c (pass_expand::execute): Do not call the
21111 outlining_inline_function hook here.
21112 * lto-wrapper.c (debug_obj): New global.
21113 (tool_cleanup): Unlink it if required.
21114 (debug_objcopy): New function.
21115 (run_gcc): Handle early debug sections in the IL files by
21116 extracting them to separate files, partially linkin them and
21117 feeding the result back as result to the linker.
21118 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
21119 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
21120 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
21121 sections into a separate segment.
21122 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
21124 (darwin_asm_dwarf_section): Likewise.
21125 (darwin_asm_output_dwarf_offset): Likewise.
21126 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
21128 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21129 Alan Hayward <alan.hayward@arm.com>
21130 David Sherwood <david.sherwood@arm.com>
21132 * read-md.h (md_reader::record_potential_iterator_use): Replace
21133 pointer argument with an rtx and an index.
21134 * read-rtl.c (iterator_group::apply_iterator): Likewise.
21135 (apply_mode_iterator): Likewise.
21136 (apply_code_iterator): Likewise.
21137 (apply_int_iterator): Likewise.
21138 (apply_subst_iterator): Likewise.
21139 (record_iterator_use): Likewise.
21140 (record_attribute_use): Likewise.
21141 (md_reader::record_potential_iterator_use): Likewise. Update calls
21142 to record_iterator_use and apply_iterator.
21143 (iterator_use): Replace ptr with x and index.
21144 (attribute_use): Likewise.
21145 (apply_attribute_uses): Update calls to apply_iterator.
21146 (apply_iterators): Likewise. Update initialization of iterator_use.
21147 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
21148 and record_potential_iterator_use.
21149 (rtx_reader::read_rtx_operand): Likewise.
21151 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
21152 Alan Hayward <alan.hayward@arm.com>
21153 David Sherwood <david.sherwood@arm.com>
21155 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
21158 2017-08-21 Richard Biener <rguenther@suse.de>
21160 PR middle-end/81884
21161 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
21162 at struct end conservatively when comparing common bases.
21164 2017-08-21 Richard Biener <rguenther@suse.de>
21166 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
21167 (mem_ref_in_stmt): Remove.
21168 (determine_max_movement): Use ref index to get at the reference.
21169 (invariantness_dom_walker::before_dom_children): Deal with
21170 lim data already initialized.
21171 (gather_mem_refs_stmt): Initialize lim data and record ref index.
21173 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21175 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
21176 (TARGET_ISA_ROUND): Ditto.
21177 (TARGET_ROUND): Ditto.
21178 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
21179 * config/i386/i386.md: Ditto.
21180 * config/i386/sse.md: Ditto.
21181 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
21182 with OPTION_MASK_ISA_SSE4_1.
21184 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
21187 * doc/extend.texi (x86 Built-in Functions): Correct the name of
21188 __builtin_ia32_lzcnt_u16.
21190 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
21193 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
21194 (rs6000_set_current_function): Rewrite function to use it.
21196 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
21199 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
21200 and TYPE_WARN_IF_NOT_ALIGN.
21201 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
21202 (handle_warn_if_not_align): New.
21203 (place_union_field): Call handle_warn_if_not_align.
21204 (place_field): Call handle_warn_if_not_align.
21205 Copy TYPE_WARN_IF_NOT_ALIGN.
21206 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
21207 (layout_type): Likewise.
21208 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
21210 (tree_decl_common): Add warn_if_not_align.
21211 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
21212 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
21213 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
21214 (DECL_WARN_IF_NOT_ALIGN): Likewise.
21215 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
21216 * doc/extend.texi: Document warn_if_not_aligned attribute.
21217 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
21219 2017-08-17 Martin Liska <mliska@suse.cz>
21222 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
21223 (get_data_dependence): Use it as pointer type.
21224 (distribute_loop): Likewise.
21226 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21228 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
21229 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
21230 (p8_vmrgow_<mode>_direct): New define_insn.
21231 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
21232 handle endianness for vmrgew and vmrgow permute patterns.
21234 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21236 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
21237 * config/rs6000/rs6000-cpus.def: Remove comment.
21238 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
21239 (POWERPC_MASKS): Likewise.
21240 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
21241 use of TARGET_VSX_TIMODE.
21242 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
21243 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
21244 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
21245 (rs6000_option_override_internal): Remove dead code.
21246 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
21247 (rs6000_legitimize_reload_address): Likewise.
21248 (rs6000_legitimate_address_p): Likewise.
21249 (rs6000_opt_masks): Delete "vsx-timode".
21250 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
21251 from function comment.
21252 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
21253 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
21254 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
21256 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
21257 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
21258 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
21259 (VSX_TI): Likewise.
21261 (define_peephole2): Likewise.
21263 2017-08-17 Martin Sebor <msebor@redhat.com>
21266 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
21267 past the end of an array.
21268 (test_pp_format): Add test cases.
21270 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
21272 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
21273 missing ECF_NOTHROW flags.
21275 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
21278 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
21279 operands residing in integer registers.
21280 (*vsx_le_perm_load_<mode>): Likewise.
21281 (*vsx_le_perm_store_<mode>): Likewise.
21282 (define_peephole2): Add peepholes to optimize the above.
21284 2017-08-17 Marek Polacek <polacek@redhat.com>
21286 PR middle-end/81814
21287 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
21288 to mimic what shorten_compare did. Change the return type to bool.
21289 (fold_cond_expr_with_comparison): Update call to
21290 operand_equal_for_comparison_p.
21291 (fold_ternary_loc): Likewise.
21293 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
21295 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
21297 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
21299 2017-08-17 Richard Biener <rguenther@suse.de>
21301 * tree-ssa-structalias.c (solve_graph): When propagating
21302 to successors update the graphs succ edges and avoid duplicate work.
21304 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
21307 * config/i386/i386.c (ix86_option_override_internal): Save target
21308 specific options after ix86_stack_protector_guard_reg was changed.
21310 2017-08-17 Richard Biener <rguenther@suse.de>
21312 PR tree-optimization/81827
21313 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
21314 (new_var_info): Initialize it conservatively.
21315 (get_call_vi): Mark register vars.
21316 (new_scalar_tmp_constraint_exp): Likewise.
21317 (handle_rhs_call): Likewise.
21318 (handle_const_call): Likewise.
21319 (create_function_info_for): Likewise.
21320 (solve_constraints): Sort varinfos to separate register from
21321 non-register vars to pack points-to solution bitmaps during
21324 2017-08-17 Marek Polacek <polacek@redhat.com>
21326 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
21328 2017-08-17 Richard Biener <rguenther@suse.de>
21330 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
21331 to true when overflow is undefined and we saturated the result.
21333 2017-08-17 Alan Modra <amodra@gmail.com>
21336 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
21337 Don't use store multiple if only one reg needs saving.
21338 (interesting_frame_related_regno): New function.
21339 (rs6000_frame_related): Don't emit frame info for regs that
21341 (rs6000_emit_epilogue): Likewise.
21343 2017-08-16 Nathan Sidwell <nathan@acm.org>
21345 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
21346 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
21347 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
21348 * tree.c (free_lang_data_in_type): Use else-if chain. Always
21349 clear TYPE_LANG_1. Remove obsolete member-function stripping.
21350 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
21351 (verify_type): Adjust for TYPE_BINFO move.
21352 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
21353 process TYPE_BINFO directly.
21354 (hash_tree): Likewise.
21355 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
21357 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
21360 2017-08-16 David Malcolm <dmalcolm@redhat.com>
21362 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
21364 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
21367 * config/i386/i386.md (*anddi_1_btr): Change predicates of
21368 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
21369 Add ix86_binary_operator_ok to insn constraint.
21370 (*iordi_1_bts): Ditto.
21371 (*xordi_1_btc): Ditto.
21372 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
21373 Update corresponding peephole2 pattern.
21377 2017-08-16 Bin Cheng <bin.cheng@arm.com>
21379 PR tree-optimization/81832
21380 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
21381 copy loop header which has IFN_LOOP_DIST_ALIAS call.
21383 2017-08-16 Marek Polacek <polacek@redhat.com>
21386 * fold-const.c (fold_indirect_ref_1): Restore original behavior
21387 regarding size_zero_node.
21389 2017-08-16 Martin Liska <mliska@suse.cz>
21392 * config.gcc: Respect previously set extra_objs in case
21395 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
21397 PR tree-optimization/81835
21398 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
21399 the phi SSA_NAME. Check that the condition in a COND_EXPR does
21400 not depend on the phi.
21402 2017-08-16 Alan Modra <amodra@gmail.com>
21404 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
21407 2017-08-16 Alan Modra <amodra@gmail.com>
21409 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
21410 (save_reg_p): ..into this. Update all callers.
21411 (first_reg_to_save): Simplify.
21413 2017-08-16 Alan Modra <amodra@gmail.com>
21415 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
21418 2017-08-15 Joseph Myers <joseph@codesourcery.com>
21422 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
21423 constant count if that count is less than 32.
21425 2017-08-15 Nathan Sidwell <nathan@acm.org>
21427 * gcc.c (execute): Emit friendlier message if inferior is killed
21428 by an external cause.
21430 2017-08-15 Richard Biener <rguenther@suse.de>
21432 PR tree-optimization/81790
21433 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
21434 CONSTRUCTORs from simplifying and VN.
21436 2017-08-14 Martin Sebor <msebor@redhat.com>
21438 * builtin-attrs.def: Add comments.
21440 2017-08-14 Martin Sebor <msebor@redhat.com>
21443 * doc/extend.texi (attribute nonstring): Document new attribute.
21445 2017-08-14 Martin Sebor <msebor@redhat.com>
21448 * tree-diagnostic.c (default_tree_printer): Handle %G.
21449 * gimple-pretty-print.h (percent_G_format): Declare new function.
21450 * gimple-pretty-print.c (percent_G_format): Define.
21451 * tree-pretty-print.c (percent_K_format): Add argument.
21453 2017-08-14 Martin Sebor <msebor@redhat.com>
21455 PR translation/79998
21456 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
21457 Remove a stray space.
21459 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
21462 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
21463 (*iordi_1_bts): Ditto.
21464 (*xordi_1_btc): Ditto.
21466 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21469 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
21470 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
21472 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
21473 quoted strings, and make more translator-friendly.
21474 (darwin_rs6000_override_options): Likewise.
21475 (rs6000_option_override_internal): Likewise.
21476 (rs6000_return_in_memory): Fix overlong line.
21477 (init_cmulative_args): Use quoted strings, and make more
21478 translator-friendly.
21479 (rs6000_pass_by_reference): Fix overlong line.
21480 (def_builtin): Use quoted strings.
21481 (altivec_expand_predicate_builtin): Use quoted strings, and make
21482 more translator-friendly.
21483 (htm_expand_builtin): Use quoted strings.
21484 (cpu_expand_builtin): Use quoted strings, and make more
21485 translator-friendly.
21486 (altivec_expand_builtin): Likewise.
21487 (paired_expand_predicate_builtin): Likewise.
21488 (rs6000_invalid_builtin): Likewise.
21489 (builtin_function_type): Use quoted strings.
21490 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
21491 more translator-friendly.
21492 (rs6000_trampoline_init): Likewise.
21493 (rs6000_handle_altivec_attribute): Likewise.
21494 (rs6000_inner_target_options): Use quoted strings.
21495 (rs6000_disable_incompatible_switches): Likewise.
21496 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
21497 strings, and make more translator-friendly.
21498 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
21500 2017-08-14 Bin Cheng <bin.cheng@arm.com>
21502 PR tree-optimization/81799
21503 * tree-loop-distribution.c (version_loop_by_alias_check): Force
21504 cond_expr to simple gimple operand.
21506 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
21508 PR middle-end/46932
21509 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
21511 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
21515 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
21516 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
21517 TARGET_GASISR_PROLOGUES.
21518 * config/avr/avr.c (avr_option_override): Same.
21519 (avr_pass_pre_proep::execute): Same.
21521 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
21524 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
21525 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
21527 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
21529 * config/i386/i386.md (*load_tp_<mode>): Redefine as
21530 define_insn_and_split. Split to a memory load from 0 in
21531 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
21532 using PTR mode iterator.
21533 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
21534 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
21535 (*add_tp_<mode>): Redefine as define_insn_and_split.
21536 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
21537 address space. Merge with *add_tp_x32 using PTR mode iterator.
21538 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
21539 Split to an add with a memory load from 0 in
21540 DEFAULT_TLS_SEG_REG address space.
21542 2017-08-12 Andrew Pinski <apinski@cavium.com>
21544 * config/aarch64/aarch64-option-extensions.def (rdma):
21545 Fix feature string to what Linux prints out in /proc/cpuinfo.
21547 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
21550 * dwarf2out.c (modified_type_die): For C typedef types that have
21551 an ultimate origin, process the ultimate origin instead of the
21553 (gen_typedef_die): Assert that input DECLs have no ultimate
21555 (gen_type_die_with_usage): For typedef variants that have an
21556 ultimate origin, just call gen_decl_die on the original DECL.
21557 (process_scope_var): Avoid creating DIEs for local typedefs and
21558 concrete static variables.
21560 2017-08-12 Alan Modra <amodra@gmail.com>
21564 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
21565 match gnu-user.h startfile.
21566 (ENDFILE_LINUX_SPEC): Similarly.
21568 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
21571 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
21573 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
21575 2017-08-11 Tamar Christina <tamar.christina@arm.com>
21576 * config/aarch64/aarch64.md (mov<mode>): Change.
21577 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
21578 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
21579 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
21581 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
21583 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
21584 for storage order barriers.
21586 2017-08-11 Martin Liska <mliska@suse.cz>
21589 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
21590 variables of void type.
21592 2017-08-11 Martin Liska <mliska@suse.cz>
21594 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
21595 TARGET_SUPPORTS_ALIASES.
21596 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
21597 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
21598 (optimize_weakref): Likewise.
21599 * symtab.c (symtab_node::noninterposable_alias): Likewise.
21600 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
21601 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
21603 2017-08-11 Martin Liska <mliska@suse.cz>
21606 * config/i386/i386.c (make_resolver_func): Do complete
21607 refactoring of the function.
21609 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21612 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
21613 * config/i386/i386.c (ix86_stack_protect_guard): Use
21614 ix86_stack_protect_guard_symbol_str to generate varible declaration.
21615 * doc/invoke.texi (x86 Options): Document
21616 -mstack-protector-guard-symbol= option.
21618 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
21620 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
21621 * config/i386/i386.c (ix86_split_stack_guard): New function.
21622 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
21623 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
21624 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
21625 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
21626 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
21627 (split_stack_space_check): Call ix86_split_stack_guard.
21629 2017-08-10 Martin Sebor <msebor@redhat.com>
21631 * print-tree.c (print_node): Print location using the established
21633 Replace spaces with colons.
21634 (debug_raw, debug): Ditto.
21636 2017-08-10 Martin Sebor <msebor@redhat.com>
21639 * pretty-print.c (pp_format): Correct the handling of %s precision.
21641 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
21644 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
21646 (ix86_finalize_stack_frame_flags): This. Also clear
21647 frame_pointer_needed if -fno-omit-frame-pointer is used without
21649 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
21650 with ix86_finalize_stack_frame_flags.
21651 (ix86_expand_epilogue): Likewise.
21652 (ix86_expand_split_stack_prologue): Likewise.
21653 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
21655 2017-08-10 Martin Liska <mliska@suse.cz>
21658 * c-attribs.c (handle_target_attribute):
21659 Report warning for an empty string argument of target attribute.
21661 2017-08-09 Jakub Jelinek <jakub@redhat.com>
21664 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
21666 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
21667 or DECL_NONLOCAL labels.
21668 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
21669 or DECL_NONLOCAL labels here.
21671 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
21673 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
21674 to indicate when early gimple folding has been disabled.
21675 (rs6000_gimple_fold_builtin): Add debug content.
21676 (rs6000_invalid_builtin): Fix whitespace.
21677 (rs6000_expand_builtin): Fix whitespace.
21678 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
21680 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
21683 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
21684 SAVE_MULTIPLE if not all the registers that saves, should be saved.
21686 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
21688 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
21689 (qdf24xx): Likewise.
21690 * config/aarch64/aarch64.md: Include falkor.md.
21691 * config/aarch64/falkor.md: New.
21693 2017-08-09 Marek Polacek <polacek@redhat.com>
21696 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
21697 * diagnostic.c (emit_diagnostic): Add a comment.
21698 (emit_diagnostic_valist): New function.
21700 2017-08-09 Marek Polacek <polacek@redhat.com>
21703 * input.c (make_location): New overload.
21704 * input.h (make_location): Declare.
21706 2017-08-08 Alan Modra <amodra@gmail.com>
21707 H.J. Lu <hongjiu.lu@intel.com>
21710 * gcc.c (NO_PIE_SPEC): Delete.
21711 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
21713 (LINK_PIE_SPEC): ..to here.
21714 (LINK_COMMAND_SPEC): Support -no-pie.
21715 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
21716 chain of crtbegin*.o selection, update for PIE_SPEC changes and
21718 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
21719 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
21720 (ENDFILE_CRTEND_SPEC): Similarly.
21722 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
21725 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
21726 (mstack-protector-guard-offset=): Ditto.
21727 * config/i386/i386.c (ix86_option_override): Handle
21728 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
21730 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
21731 ix86_stack_protect_guard_offset variables.
21732 (TARGET_STACK_PROTECT_GUARD): Always define.
21733 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
21734 and -mstack-protector-guard-offset= options.
21736 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21738 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
21739 boundary case for the last candidate.
21741 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21743 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
21744 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
21746 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21748 PR middle-end/19706
21749 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
21750 * config/aarch64/aarch64-builtins.c
21751 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
21752 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
21753 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
21755 2017-08-08 Tamar Christina <tamar.christina@arm.com>
21756 Andrew Pinski <pinskia@gmail.com>
21758 PR middle-end/19706
21759 * internal-fn.def (XORSIGN): New.
21760 * optabs.def (xorsign_optab): New.
21761 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
21762 (convert_expand_mult_copysign): New.
21763 (pass_optimize_widening_mul::execute): Call
21764 convert_expand_mult_copysign.
21766 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21768 PR tree-optimization/81354
21769 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
21770 Insert on edges rather than explicitly creating landing pads.
21771 (analyze_candidates_and_replace): Commit edge inserts.
21773 2017-08-08 Richard Biener <rguenther@suse.de>
21775 PR middle-end/81719
21776 * tree-ssa-loop-niter.c: Include tree-dfa.h.
21777 (expand_simple_operations): Also look through ADDR_EXPRs with
21778 MEM_REF bases treating them as POINTER_PLUS_EXPR.
21780 2017-08-08 Richard Biener <rguenther@suse.de>
21782 PR tree-optimization/81723
21783 * tree-vect-slp.c (struct bst_traits): New hash traits.
21784 (bst_fail): New global.
21785 (vect_build_slp_tree_2): New worker, split out from ...
21786 (vect_build_slp_tree): ... this now wrapping it with using
21787 bst_fail set to cache SLP tree build fails. Properly handle
21789 (vect_analyze_slp_instance): Allocate and free bst_fail.
21791 2017-08-08 Martin Liska <mliska@suse.cz>
21794 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
21795 LABEL_DECLs that can be from a different function.
21797 2017-08-08 Bin Cheng <bin.cheng@arm.com>
21799 PR tree-optimization/81744
21800 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
21801 loop's number of iterations.
21803 2017-08-08 Martin Liska <mliska@suse.cz>
21805 * asan.c: Include header files.
21806 * attribs.c (build_decl_attribute_variant): New function moved
21808 (build_type_attribute_qual_variant): Likewise.
21809 (cmp_attrib_identifiers): Likewise.
21810 (simple_cst_list_equal): Likewise.
21811 (omp_declare_simd_clauses_equal): Likewise.
21812 (attribute_value_equal): Likewise.
21813 (comp_type_attributes): Likewise.
21814 (build_type_attribute_variant): Likewise.
21815 (lookup_ident_attribute): Likewise.
21816 (remove_attribute): Likewise.
21817 (merge_attributes): Likewise.
21818 (merge_type_attributes): Likewise.
21819 (merge_decl_attributes): Likewise.
21820 (merge_dllimport_decl_attributes): Likewise.
21821 (handle_dll_attribute): Likewise.
21822 (attribute_list_equal): Likewise.
21823 (attribute_list_contained): Likewise.
21824 * attribs.h (lookup_attribute): New function moved from tree.[ch].
21825 (lookup_attribute_by_prefix): Likewise.
21826 * bb-reorder.c: Include header files.
21827 * builtins.c: Likewise.
21828 * calls.c: Likewise.
21829 * cfgexpand.c: Likewise.
21830 * cgraph.c: Likewise.
21831 * cgraphunit.c: Likewise.
21832 * convert.c: Likewise.
21833 * dwarf2out.c: Likewise.
21834 * final.c: Likewise.
21835 * fold-const.c: Likewise.
21836 * function.c: Likewise.
21837 * gimple-expr.c: Likewise.
21838 * gimple-fold.c: Likewise.
21839 * gimple-pretty-print.c: Likewise.
21840 * gimple.c: Likewise.
21841 * gimplify.c: Likewise.
21842 * hsa-common.c: Likewise.
21843 * hsa-gen.c: Likewise.
21844 * internal-fn.c: Likewise.
21845 * ipa-chkp.c: Likewise.
21846 * ipa-cp.c: Likewise.
21847 * ipa-devirt.c: Likewise.
21848 * ipa-fnsummary.c: Likewise.
21849 * ipa-inline.c: Likewise.
21850 * ipa-visibility.c: Likewise.
21852 * lto-cgraph.c: Likewise.
21853 * omp-expand.c: Likewise.
21854 * omp-general.c: Likewise.
21855 * omp-low.c: Likewise.
21856 * omp-offload.c: Likewise.
21857 * omp-simd-clone.c: Likewise.
21858 * opts-global.c: Likewise.
21859 * passes.c: Likewise.
21860 * predict.c: Likewise.
21861 * sancov.c: Likewise.
21862 * sanopt.c: Likewise.
21863 * symtab.c: Likewise.
21864 * toplev.c: Likewise.
21865 * trans-mem.c: Likewise.
21866 * tree-chkp.c: Likewise.
21867 * tree-eh.c: Likewise.
21868 * tree-into-ssa.c: Likewise.
21869 * tree-object-size.c: Likewise.
21870 * tree-parloops.c: Likewise.
21871 * tree-profile.c: Likewise.
21872 * tree-ssa-ccp.c: Likewise.
21873 * tree-ssa-live.c: Likewise.
21874 * tree-ssa-loop.c: Likewise.
21875 * tree-ssa-sccvn.c: Likewise.
21876 * tree-ssa-structalias.c: Likewise.
21877 * tree-ssa.c: Likewise.
21878 * tree-streamer-in.c: Likewise.
21879 * tree-vectorizer.c: Likewise.
21880 * tree-vrp.c: Likewise.
21881 * tsan.c: Likewise.
21882 * ubsan.c: Likewise.
21883 * varasm.c: Likewise.
21884 * varpool.c: Likewise.
21885 * tree.c: Remove functions moved to attribs.[ch].
21886 * tree.h: Likewise.
21887 * config/aarch64/aarch64.c: Add attrs.h header file.
21888 * config/alpha/alpha.c: Likewise.
21889 * config/arc/arc.c: Likewise.
21890 * config/arm/arm.c: Likewise.
21891 * config/avr/avr.c: Likewise.
21892 * config/bfin/bfin.c: Likewise.
21893 * config/c6x/c6x.c: Likewise.
21894 * config/cr16/cr16.c: Likewise.
21895 * config/cris/cris.c: Likewise.
21896 * config/darwin.c: Likewise.
21897 * config/epiphany/epiphany.c: Likewise.
21898 * config/fr30/fr30.c: Likewise.
21899 * config/frv/frv.c: Likewise.
21900 * config/ft32/ft32.c: Likewise.
21901 * config/h8300/h8300.c: Likewise.
21902 * config/i386/winnt.c: Likewise.
21903 * config/ia64/ia64.c: Likewise.
21904 * config/iq2000/iq2000.c: Likewise.
21905 * config/lm32/lm32.c: Likewise.
21906 * config/m32c/m32c.c: Likewise.
21907 * config/m32r/m32r.c: Likewise.
21908 * config/m68k/m68k.c: Likewise.
21909 * config/mcore/mcore.c: Likewise.
21910 * config/microblaze/microblaze.c: Likewise.
21911 * config/mips/mips.c: Likewise.
21912 * config/mmix/mmix.c: Likewise.
21913 * config/mn10300/mn10300.c: Likewise.
21914 * config/moxie/moxie.c: Likewise.
21915 * config/msp430/msp430.c: Likewise.
21916 * config/nds32/nds32-isr.c: Likewise.
21917 * config/nds32/nds32.c: Likewise.
21918 * config/nios2/nios2.c: Likewise.
21919 * config/nvptx/nvptx.c: Likewise.
21920 * config/pa/pa.c: Likewise.
21921 * config/pdp11/pdp11.c: Likewise.
21922 * config/powerpcspe/powerpcspe.c: Likewise.
21923 * config/riscv/riscv.c: Likewise.
21924 * config/rl78/rl78.c: Likewise.
21925 * config/rx/rx.c: Likewise.
21926 * config/s390/s390.c: Likewise.
21927 * config/sh/sh.c: Likewise.
21928 * config/sol2.c: Likewise.
21929 * config/sparc/sparc.c: Likewise.
21930 * config/spu/spu.c: Likewise.
21931 * config/stormy16/stormy16.c: Likewise.
21932 * config/tilegx/tilegx.c: Likewise.
21933 * config/tilepro/tilepro.c: Likewise.
21934 * config/v850/v850.c: Likewise.
21935 * config/vax/vax.c: Likewise.
21936 * config/visium/visium.c: Likewise.
21937 * config/xtensa/xtensa.c: Likewise.
21939 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
21942 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
21943 constraints since the -mupper-regs-* switches have been
21945 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
21946 into a vector from a double word element that was extracted from
21947 another vector, and eliminate extra XXPERMDI instructions.
21948 (vsx_concat_<mode>_2): Likewise.
21949 (vsx_concat_<mode>_3): Likewise.
21950 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
21951 concat to allow optimizing inserts from previous extracts.
21953 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
21955 * config/i386/i386.c (ix86_stack_protect_guard): Generate
21956 memory reference to a SSP offset in TLS address space.
21957 (ix86_print_operand) <case '@'>: Remove.
21958 (ix86_print_operand_punct_valid_p): Remove '@' code.
21959 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
21960 UNSPEC_SP_TLS_TEST.
21961 (stack_tls_protect_set_<mode>): Remove.
21962 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
21963 (stack_tls_protect_test_<mode>): Remove.
21964 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
21966 2017-08-07 Olivier Hainque <hainque@adacore.com>
21969 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
21971 2017-08-07 Douglas Rupp <rupp@adacore.com>
21973 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
21974 variable was referenced as multidir in command.
21976 2017-08-07 Jakub Jelinek <jakub@redhat.com>
21979 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
21982 2017-08-07 Martin Liska <mliska@suse.cz>
21984 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
21985 * config/rl78/rl78.c: Add include of attribs.h.
21986 * config/sh/sh.c: Likewise.
21987 * config/v850/v850.c: Likewise.
21989 2017-08-07 Tom de Vries <tom@codesourcery.com>
21991 PR middle-end/78266
21992 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
21994 2017-08-07 Martin Liska <mliska@suse.cz>
21996 * config/mips/mips.c: Include attribs.h.
21998 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
22001 * doc/invoke.texi: Document change in behvaior for -Ofast for
22004 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
22006 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
22008 (aarch64_pop_regs): Likewise.
22009 (aarch64_gen_load_pair): Likewise.
22010 (aarch64_save_callee_saves): Likewise.
22011 (aarch64_restore_callee_saves): Likewise.
22013 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22015 * config/i386/i386.c: Revert the last change.
22017 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22020 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
22022 (ix86_finalize_stack_frame_flags): This. Also clear
22023 frame_pointer_needed if -fno-omit-frame-pointer is used without
22025 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
22026 with ix86_finalize_stack_frame_flags.
22027 (ix86_expand_epilogue): Likewise.
22028 (ix86_expand_split_stack_prologue): Likewise.
22030 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
22033 * config/i386/i386.c (get_builtin_code_for_version): Set priority
22034 to P_AES for Westmere.
22036 2017-08-07 Jonathan Yong <10walls@gmail.com>
22038 * config/i386/mingw.opt (fset-stack-executable): Removed.
22039 * config/i386/cygming.opt (fset-stack-executable): Moved
22041 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
22043 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
22045 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
22047 2017-08-07 Marek Polacek <polacek@redhat.com>
22049 PR middle-end/81737
22050 * fold-const.c (fold_indirect_ref_1): Check type_domain.
22052 2017-08-07 Martin Liska <mliska@suse.cz>
22054 * attribs.h (canonicalize_attr_name): New function.
22055 (cmp_attribs): Move from c-format.c and adjusted.
22056 (is_attribute_p): Moved from tree.h.
22057 * tree-inline.c: Add new includes.
22058 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
22059 (private_is_attribute_p): Remove.
22060 (private_lookup_attribute): Likewise.
22061 (private_lookup_attribute_by_prefix): Simplify.
22062 (remove_attribute): Use is_attribute_p.
22063 * tree.h: Remove removed declarations.
22065 2017-08-07 Jakub Jelinek <jakub@redhat.com>
22067 PR middle-end/81698
22068 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
22069 instead of computing it in the function. Formatting fix.
22070 (expand_case): Don't rely on default_edge being the first edge,
22071 clear it if removing it, pass default_edge to
22072 emit_case_dispatch_table.
22073 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
22076 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
22078 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
22079 insn in the function, emit NOP after the insn.
22081 2017-08-06 Tom de Vries <tom@codesourcery.com>
22083 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
22086 2017-08-06 Tom de Vries <tom@codesourcery.com>
22088 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
22091 2017-08-04 Yury Gribov <tetra2005@gmail.com>
22093 PR tree-optimization/57371
22094 * match.pd: New pattern.
22096 2017-08-04 Marek Polacek <polacek@redhat.com>
22098 PR middle-end/81695
22099 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
22100 perform the computation in offset_int.
22102 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22104 PR tree-optimization/81136
22105 * tree-vectorizer.h: Include tree-hash-traits.h.
22106 (vec_base_alignments): New typedef.
22107 (vec_info): Add a base_alignments field.
22108 (vect_record_base_alignments): Declare.
22109 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
22111 (DR_IS_CONDITIONAL_IN_STMT): New macro.
22112 (create_data_ref): Add an is_conditional_in_stmt argument.
22113 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
22114 the is_conditional_in_stmt field.
22115 (data_ref_loc): Add an is_conditional_in_stmt field.
22116 (get_references_in_stmt): Set the is_conditional_in_stmt field.
22117 (find_data_references_in_stmt): Update call to create_data_ref.
22118 (graphite_find_data_references_in_stmt): Likewise.
22119 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
22120 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
22121 (vect_record_base_alignment): New function.
22122 (vect_record_base_alignments): Likewise.
22123 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
22124 for nested statements even if we fail to compute a misalignment.
22125 Use pooled base alignments for unconditional references.
22126 (vect_find_same_alignment_drs): Compare base addresses instead
22128 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
22129 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
22131 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22133 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
22134 Add an explicit name for the enum. Use auto_vec for slp_instances
22135 and grouped_stores.
22136 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
22138 (_bb_vec_info): Add a constructor and destructor.
22139 (vinfo_for_stmt): Return NULL for uids of -1 as well.
22140 (destroy_loop_vec_info): Delete.
22141 (vect_destroy_datarefs): Likewise.
22142 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
22143 (vec_info::vec_info): New function.
22144 (vec_info::~vec_info): Likewise.
22145 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
22146 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
22147 destroy_loop_vec_info.
22148 * tree-vect-loop.c (new_loop_vec_info): Replace with...
22149 (_loop_vec_info::_loop_vec_info): ...this.
22150 (destroy_loop_vec_info): Replace with...
22151 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
22152 the stmt_vec_infos. Leave handling of vec_info information to its
22153 destructor. Remove explicit vector releases.
22154 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
22155 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
22156 * tree-vect-slp.c (new_bb_vec_info): Replace with...
22157 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
22158 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
22159 (destroy_bb_vec_info): Replace with...
22160 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
22161 information to its destructor.
22162 (vect_slp_analyze_bb_1): Use new and delete instead of
22163 new_bb_vec_info and destroy_bb_vec_info.
22164 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
22167 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
22169 * tree-data-ref.h (subscript): Add access_fn field.
22170 (data_dependence_relation): Add could_be_independent_p.
22171 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
22172 (same_access_functions): Move to tree-data-ref.c.
22173 * tree-data-ref.c (ref_contains_union_access_p): New function.
22174 (access_fn_component_p): Likewise.
22175 (access_fn_components_comparable_p): Likewise.
22176 (dr_analyze_indices): Add a reference to access_fn_component_p.
22177 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
22179 (constant_access_functions): Likewise.
22180 (add_other_self_distances): Likewise.
22181 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
22182 (initialize_data_dependence_relation): Use XCNEW and remove
22183 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
22184 of access functions that have the same type. Allow the
22185 subsequence to end with different bases in some circumstances.
22186 Record the chosen access functions in SUB_ACCESS_FN.
22187 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
22188 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
22189 (subscript_dependence_tester_1): Likewise dra and drb.
22190 (build_classic_dist_vector): Update calls accordingly.
22191 (subscript_dependence_tester): Likewise.
22192 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
22193 DDR_COULD_BE_INDEPENDENT_P.
22194 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
22195 comp_alias_ddrs instead of may_alias_ddrs.
22196 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
22198 (vect_analyze_data_ref_dependence): Use it if
22199 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
22200 distance vectors if that fails.
22201 (dependence_distance_ge_vf): New function.
22202 (vect_prune_runtime_alias_test_list): Use it. Don't clear
22203 LOOP_VINFO_MAY_ALIAS_DDRS.
22205 2017-08-04 Richard Biener <rguenther@suse.de>
22207 PR middle-end/81705
22208 * fold-const.c (fold_binary_loc): Properly restrict
22209 minus_var0 && minus_var1 case when associating undefined overflow
22212 2017-08-04 Tom de Vries <tom@codesourcery.com>
22214 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
22216 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22218 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22219 Don't start diagnostic messages with a capital letter.
22220 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22222 (rs6000_invalid_builtin): Likewise.
22223 (rs6000_trampoline_init): Likewise.
22225 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22228 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
22229 after setting changeable df flags.
22231 2017-08-03 Richard Biener <rguenther@suse.de>
22233 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
22234 up if the use is in USE - X.
22236 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
22238 * toplev.c (dumpfile.h): New include.
22239 (internal_error_reentered): New static function. Use it...
22240 (internal_error_function): ...here to handle reentered internal_error.
22242 2017-08-03 Richard Biener <rguenther@suse.de>
22244 PR middle-end/81148
22245 * fold-const.c (split_tree): Add minus_var and minus_con
22246 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
22247 here but always use minus_*.
22248 (associate_trees): Assert we never associate with MINUS_EXPR
22249 and NULL first operand. Do not recurse for PLUS_EXPR operands
22250 when associating as MINUS_EXPR either.
22251 (fold_binary_loc): Track minus_var and minus_con.
22253 2017-08-03 Tom de Vries <tom@codesourcery.com>
22256 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
22257 ACCEL_COMPILER, apply finish_options on
22258 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
22260 2017-08-03 Tom de Vries <tom@codesourcery.com>
22263 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
22264 function_entry_patch_area_size > 0.
22266 2017-08-03 Jakub Jelinek <jakub@redhat.com>
22269 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
22270 instead of 10??LU, perform unit multiplication in wide_int,
22271 don't change alloc_object_size_limit if the limit is larger
22274 PR tree-optimization/81655
22275 PR tree-optimization/81588
22276 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
22277 the case when ranges[i].low and high are 1 for unsigned type with
22280 PR middle-end/81052
22281 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
22282 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
22284 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22286 * tree-vrp.h: Add include guard.
22288 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
22291 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
22292 (ud2): New insn pattern.
22293 * config/i386/i386.c (ix86_expand_epilogue):
22294 For naked functions, generate ud2 instead of trap insn.
22296 2017-08-02 Marek Polacek <polacek@redhat.com>
22299 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
22301 2017-08-02 Tom de Vries <tom@codesourcery.com>
22302 Cesar Philippidis <cesar@codesourcery.com>
22304 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
22305 Add missing edge probabilities.
22307 2017-08-02 Tamar Christina <tamar.christina@arm.com>
22309 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
22310 Correct endianness.
22312 2017-08-02 Jakub Jelinek <jakub@redhat.com>
22314 PR middle-end/79499
22315 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
22316 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
22317 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
22318 prologue_seq sequences - if any.
22320 2017-08-02 Richard Biener <rguenther@suse.de>
22322 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
22323 via vectors if supported, integer extracts via punning if supported
22324 or otherwise vector extracts.
22326 2017-08-02 Richard Biener <rguenther@suse.de>
22328 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
22330 (bitmap_insert_into_set): ... this.
22332 2017-08-02 Richard Biener <rguenther@suse.de>
22334 PR tree-optimization/81633
22336 2015-08-17 Alan Hayward <alan.hayward@arm.com>
22338 PR tree-optimization/71752
22339 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
22341 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
22343 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
22344 (machine_function::call_ms2sysv_pad_out): Remove field.
22345 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
22346 (ix86_compute_frame_layout): Likewise.
22348 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
22351 * config/i386/i386.c (ix86_set_func_type): Disallow naked
22352 attribute with interrupt attribute.
22354 2017-08-01 Andrew Pinski <apinski@cavium.com>
22356 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
22357 BIT_INSERT_EXPR's operand 1
22358 to see if the types precision matches.
22360 2017-08-01 Martin Liska <mliska@suse.cz>
22362 PR middle-end/70140
22363 * builtins.c (expand_builtin_memcpy_args): Remove.
22364 (expand_builtin_memcpy): Call newly added function
22365 expand_builtin_memory_copy_args.
22366 (expand_builtin_memcpy_with_bounds): Likewise.
22367 (expand_builtin_mempcpy): Remove last argument.
22368 (expand_builtin_mempcpy_with_bounds): Likewise.
22369 (expand_builtin_memory_copy_args): New function created from
22370 expand_builtin_mempcpy_args with small modifications.
22371 (expand_builtin_mempcpy_args): Remove.
22372 (expand_builtin_stpcpy): Remove unused argument.
22373 (expand_builtin): Likewise.
22374 (expand_builtin_with_bounds): Likewise.
22376 2017-08-01 Martin Liska <mliska@suse.cz>
22379 Make mempcpy more optimal (PR middle-end/70140).
22381 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22384 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
22385 __builtin_vec_cmpne verify both arguments are compatible vectors
22386 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
22387 verify arg1_type is a pointer or array type. For __builtin_vec_st,
22388 move computation of aligned to after checking the argument types.
22392 * config/rs6000/vsx.md (vextract_fp_from_shorth,
22393 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
22396 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
22397 Doug Rupp <rupp@adacore.com>
22398 Olivier Hainque <hainque@adacore.com>
22400 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
22401 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
22403 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
22404 for TARGET_OS_CPP_BUILTIN.
22405 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
22406 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
22408 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
22409 passing required abi options to the assembler for EABI configurations.
22410 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
22411 of .text.hot and .text.unlikely sections for kernel modules when
22412 using ARM style exceptions.
22413 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
22414 options. Add EXTRA_CC1_SPEC.
22415 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
22417 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
22419 (ARM_TARGET2_DWARF_FORMAT): Define.
22420 * config/arm/t-vxworks: Adjust multilib control to removal of the
22421 Diab command line options.
22423 2017-08-01 Martin Liska <mliska@suse.cz>
22425 PR gcov-profile/81561
22426 * gcov.c (unblock): Make unblocking safe as we need to preserve
22427 index correspondence of blocks and block_lists.
22429 2017-08-01 Richard Biener <rguenther@suse.de>
22431 PR tree-optimization/81181
22432 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
22433 (compute_antic): ... end of iteration here.
22435 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
22437 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
22438 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
22439 (ftree-slp-vectorize): Likewise.
22440 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
22441 can no longer be set independent of flag_tree_loop_vectorize.
22442 * omp-general.c (emp_max_vf): Likewise.
22443 * opts.c (enable_fdo_optimizations): Remove references to
22444 flag_tree_vectorize, these are now implicit.
22445 (common_handle_option): Remove handling for OPT_ftree_vectorize,
22446 and leave it for the options machinery.
22448 2017-08-01 Martin Liska <mliska@suse.cz>
22450 PR middle-end/70140
22451 * builtins.c (expand_builtin_memcpy_args): Remove.
22452 (expand_builtin_memcpy): Call newly added function
22453 expand_builtin_memory_copy_args.
22454 (expand_builtin_memcpy_with_bounds): Likewise.
22455 (expand_builtin_mempcpy): Remove last argument.
22456 (expand_builtin_mempcpy_with_bounds): Likewise.
22457 (expand_builtin_memory_copy_args): New function created from
22458 expand_builtin_mempcpy_args with small modifications.
22459 (expand_builtin_mempcpy_args): Remove.
22460 (expand_builtin_stpcpy): Remove unused argument.
22461 (expand_builtin): Likewise.
22462 (expand_builtin_with_bounds): Likewise.
22464 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22467 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
22468 print "ds:" only for immediates in generic address space.
22470 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
22473 * config/i386/i386.c (ix86_funciton_naked): New prototype.
22474 (ix86_function_ok_for_sibcall): Return false for naked functions.
22476 2017-08-01 Richard Biener <rguenther@suse.de>
22478 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
22479 (compute_antic): Seed worklist with exit block predecessors.
22480 * cfganal.c (dfs_find_deadend): For a cycle return the source
22481 of the edge closing it.
22483 2017-08-01 Tamar Christina <tamar.christina@arm.com>
22485 * config/aarch64/aarch64.c
22486 (aarch64_can_const_movi_rtx_p): Move 0 check.
22488 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22490 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
22491 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
22493 * match.pd: Ditto in address comparison pattern.
22495 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22497 PR tree-optimization/81627
22498 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
22499 closed ssa form for store-store chain.
22501 2017-08-01 Bin Cheng <bin.cheng@arm.com>
22503 PR tree-optimization/81620
22504 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
22505 for store-store chain.
22507 2017-08-01 Jakub Jelinek <jakub@redhat.com>
22509 PR tree-optimization/81588
22510 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
22511 ranges[i].in_p, invert comparison code ccode. For >/>=,
22512 swap rhs1 and rhs2 and comparison code unconditionally,
22513 for </<= don't do that. Don't swap rhs1/rhs2 again if
22514 ranges[i].in_p, instead invert comparison code ccode if
22515 opcode or oe->rank is BIT_IOR_EXPR.
22518 * optabs.def (vec_extract_optab, vec_init_optab): Change from
22519 a direct optab to conversion optab.
22520 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
22521 with GET_MODE_INNER as last argument instead of optab_handler.
22522 * expmed.c (extract_bit_field_1): Likewise. Use vector from
22523 vector extraction if possible and optab is available.
22524 * expr.c (store_constructor): Use convert_optab_handler instead
22525 of optab_handler. Use vector initialization from smaller
22526 vectors if possible and optab is available.
22527 * tree-vect-stmts.c (vectorizable_load): Likewise.
22528 * doc/md.texi (vec_extract, vec_init): Document that the optabs
22529 now have two modes.
22530 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
22531 of vec_init from half-sized vectors with the same element mode.
22532 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
22533 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
22534 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
22535 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
22536 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
22537 after mode in gen_vec_extract* calls.
22538 (vec_extract<mode>): Renamed to ...
22539 (vec_extract<mode><ssescalarmodelower>): ... this.
22540 (vec_extract<mode><ssehalfvecmodelower>): New expander.
22541 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
22542 element mode after mode in gen_vec_init* calls.
22543 (VEC_INIT_HALF_MODE): New mode iterator.
22544 (vec_init<mode>): Renamed to ...
22545 (vec_init<mode><ssescalarmodelower>): ... this.
22546 (vec_init<mode><ssehalfvecmodelower>): New expander.
22547 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
22548 (vec_extractv2sfsf): ... this.
22549 (vec_initv2sf): Renamed to ...
22550 (vec_initv2sfsf): ... this.
22551 (vec_extractv2si): Renamed to ...
22552 (vec_extractv2sisi): ... this.
22553 (vec_initv2si): Renamed to ...
22554 (vec_initv2sisi): ... this.
22555 (vec_extractv4hi): Renamed to ...
22556 (vec_extractv4hihi): ... this.
22557 (vec_initv4hi): Renamed to ...
22558 (vec_initv4hihi): ... this.
22559 (vec_extractv8qi): Renamed to ...
22560 (vec_extractv8qiqi): ... this.
22561 (vec_initv8qi): Renamed to ...
22562 (vec_initv8qiqi): ... this.
22563 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
22564 (vec_init<mode>): Renamed to ...
22565 (vec_init<mode><VEC_base_l>): ... this.
22566 (vec_extract<mode>): Renamed to ...
22567 (vec_extract<mode><VEC_base_l>): ... this.
22568 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
22569 (vec_initv2sfsf): ... this.
22570 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
22571 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22572 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
22573 element mode after mode in gen_vec_init* calls.
22574 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
22575 (vec_init<mode><Vel>): ... this.
22576 (vec_extract<mode>): Renamed to ...
22577 (vec_extract<mode><Vel>): ... this.
22578 * config/aarch64/iterators.md (Vel): New mode attribute.
22579 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
22580 Add element mode after mode in gen_vec_extract* calls.
22581 * config/s390/vector.md (non_vec_l): New mode attribute.
22582 (vec_extract<mode>): Renamed to ...
22583 (vec_extract<mode><non_vec_l>): ... this.
22584 (vec_init<mode>): Renamed to ...
22585 (vec_init<mode><non_vec_l>): ... this.
22586 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
22587 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
22589 * config/arm/iterators.md (V_elem_l): New mode attribute.
22590 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
22591 (vec_extract<mode><V_elem_l>): ... this.
22592 (vec_extractv2di): Renamed to ...
22593 (vec_extractv2didi): ... this.
22594 (vec_init<mode>): Renamed to ...
22595 (vec_init<mode><V_elem_l>): ... this.
22596 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
22597 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
22598 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
22599 Add element mode after gen_vec_extract* calls.
22600 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
22601 (vec_init<mode><unitmode>): ... this.
22602 (vec_extract<mode>): Renamed to ...
22603 (vec_extract<mode><unitmode>): ... this.
22604 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
22605 (vec_init<mode><unitmode>): ... this.
22606 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
22607 (vec_initv2sfsf): ... this.
22608 (vec_extractv2sf): Renamed to ...
22609 (vec_extractv2sfsf): ... this.
22610 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
22611 Add element mode after gen_vec_extract* calls.
22612 * config/mips/mips.md (unitmode): New mode iterator.
22613 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
22614 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
22615 * config/spu/spu.md (inner_l): New mode attribute.
22616 (vec_init<mode>): Renamed to ...
22617 (vec_init<mode><inner_l>): ... this.
22618 (vec_extract<mode>): Renamed to ...
22619 (vec_extract<mode><inner_l>): ... this.
22620 * config/sparc/sparc.md (veltmode): New mode iterator.
22621 (vec_init<VMALL:mode>): Renamed to ...
22622 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
22623 * config/ia64/vect.md (vec_initv2si): Renamed to ...
22624 (vec_initv2sisi): ... this.
22625 (vec_initv2sf): Renamed to ...
22626 (vec_initv2sfsf): ... this.
22627 (vec_extractv2sf): Renamed to ...
22628 (vec_extractv2sfsf): ... this.
22629 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
22630 (vec_init<mode>): Renamed to ...
22631 (vec_init<mode><VEC_base_l>): ... this.
22632 (vec_extract<mode>): Renamed to ...
22633 (vec_extract<mode><VEC_base_l>): ... this.
22634 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
22635 (vec_initv2sfsf): ... this.
22636 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
22637 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
22638 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
22639 gen_vec_init* calls.
22641 2017-08-01 Richard Biener <rguenther@suse.de>
22643 PR tree-optimization/81297
22644 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
22645 TREE_OVERFLOW from INTEGER_CSTs.
22647 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
22649 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
22651 2017-07-31 Carl Love <cel@us.ibm.com>
22653 * config/rs6000/rs6000-c: Add support for built-in functions
22654 vector signed char vec_xl_be (signed long long, signed char *);
22655 vector unsigned char vec_xl_be (signed long long, unsigned char *);
22656 vector signed int vec_xl_be (signed long long, signed int *);
22657 vector unsigned int vec_xl_be (signed long long, unsigned int *);
22658 vector signed long long vec_xl_be (signed long long, signed long long *);
22659 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
22660 vector signed short vec_xl_be (signed long long, signed short *);
22661 vector unsigned short vec_xl_be (signed long long, unsigned short *);
22662 vector double vec_xl_be (signed long long, double *);
22663 vector float vec_xl_be (signed long long, float *);
22664 * config/rs6000/altivec.h (vec_xl_be): Add #define.
22665 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
22666 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
22668 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
22669 (altivec_expand_builtin): Add switch statement to call
22670 altivec_expand_xl_be for each builtin.
22671 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
22672 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
22673 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
22674 __builtin_vsx_le_be_v16qi.
22675 * doc/extend.texi: Update the built-in documentation file for the
22676 new built-in functions.
22678 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22681 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
22683 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
22685 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22687 * config.gcc: Add z14.
22688 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
22689 CPU model numbers for z13s and z14.
22690 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
22692 * config/s390/s390-opts.h (enum processor_type): Rename
22693 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22694 * config/s390/s390.c (processor_table): Add field for CPU name to
22695 be passed to Binutils.
22696 (s390_asm_output_machine_for_arch): Use the new field in
22697 processor_table for Binutils.
22698 (s390_expand_builtin): Replace arch12 with z14.
22699 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
22700 (s390_get_sched_attrmask): Likewise.
22701 (s390_get_unit_mask): Likewise.
22702 * config/s390/s390.opt: Add z14 to processor_type enum.
22704 2017-07-31 Martin Jambor <mjambor@suse.cz>
22707 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
22708 regardless of optimization level.
22710 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
22711 Martin Liska <mliska@suse.cz>
22713 * predict.def: Remove old comment and adjust probability.
22714 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
22715 PREDICT statements.
22717 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
22720 * config/i386/i386.c (ix86_function_naked): New function.
22721 (ix86_can_use_return_insn_p): Return false for naked functions.
22722 (ix86_expand_prologue): Skip prologue for naked functions.
22723 (ix86_expand_epilogue): Skip epilogue for naked functions
22724 and emit trap instruction.
22725 (ix86_warn_func_return): New function.
22726 (ix86_attribute_table): Add "naked" attribute specification.
22727 (TARGET_WARN_FUNC_RETURN): Define.
22728 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
22730 2017-07-31 Martin Liska <mliska@suse.cz>
22732 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
22733 (dump_gimple_bb_header): Always dump BB info.
22734 (pp_cfg_jump): Do not append info about BB when dumping a jump.
22736 2017-07-31 Martin Liska <mliska@suse.cz>
22739 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
22740 also with current_function_decl non-null equality.
22742 2017-07-31 Jakub Jelinek <jakub@redhat.com>
22745 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
22746 change type to the element type, instead add eltype variable and
22747 use it where we are interested in the element type.
22749 PR tree-optimization/81603
22750 * ipa-polymorphic-call.c
22751 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
22752 offset arithmetic in offset_int, bail out if the resulting bit offset
22753 doesn't fit into shwi.
22755 2017-07-31 Martin Liska <mliska@suse.cz>
22757 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
22758 (gimplify_save_expr): Fix comment.
22760 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
22763 * config/i386/i386.c (ix86_function_arg): Update arguments for
22765 (ix86_compute_frame_layout): Set the initial stack offset to
22766 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
22767 INCOMING_FRAME_SP_OFFSET.
22768 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
22769 stack before exception handler returns.
22770 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
22771 the 'ERROR_CODE' for exception handler.
22773 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
22775 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
22776 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
22777 (ASM_OUTPUT_REG_POP): Ditto.
22778 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
22779 instead of asm_fprintf to output pure string.
22781 2017-07-29 Jakub Jelinek <jakub@redhat.com>
22783 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
22784 to imported_module_or_decl hook.
22785 (debug_nothing_tree_tree_tree_bool): Remove.
22786 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
22787 * debug.c (do_nothing_debug_hooks): Use
22788 debug_nothing_tree_tree_tree_bool_bool instead of
22789 debug_nothing_tree_tree_tree_bool.
22790 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
22791 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
22792 * sdbout.c (sdb_debug_hooks): Likewise.
22793 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
22794 (gen_namespace_die): Add DW_AT_export_symbols attribute if
22796 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
22797 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
22798 attribute, don't add anything.
22800 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22802 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
22803 (fold_build2_stat_loc): Likewise.
22804 (fold_build3_stat_loc): Likewise.
22805 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
22806 (fold_build1_loc): Remove macro.
22807 (fold_build2_loc): Likewise.
22808 (fold_build3_loc): Likewise.
22810 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22812 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
22813 (gimple_build_debug_bind_source_stat): Likewise.
22814 * gimple.h (gimple_build_debug_bind): Remove macro.
22815 (gimple_build_debug_bind_source): Likewise.
22817 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22819 * bitmap.c (bitmap_alloc): Adjust.
22820 (bitmap_gc_alloc): Likewise.
22821 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
22823 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22825 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
22826 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
22827 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
22828 (bitmap_gc_alloc_stat): Likewise.
22829 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
22831 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22833 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
22834 * rtl.h (shallow_copy_rtx): Remove macro.
22836 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22838 * emit-rtl.c (gen_raw_REG): Adjust.
22839 * gengenrtl.c (gendef): Likewise.
22840 * rtl.c (rtx_alloc_stat): Remove _stat from name.
22841 * rtl.h (rtx_alloc): Remove macro.
22843 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22845 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
22846 (build_tree_list_stat): Likewise.
22847 * tree.h (build_tree_list): Remove macro.
22848 (build_tree_list_vec): Likewise.
22850 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22852 * tree.c (make_vector_stat): Remove _stat from name.
22853 (build_vector_stat): Likewise.
22854 * tree.h (make_vector_stat): Remove macro.
22855 (build_vector_stat): Likewise.
22857 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22859 * tree.h (build_var_debug_value): Remove prototype.
22861 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22863 * tree.c (tree_cons_stat): Remove _stat from name.
22864 * tree.h (tree_cons): Remove macro.
22866 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22868 * tree.c (build_vl_exp_stat): Remove _stat from name.
22869 * tree.h (build_vl_exp): Remove macro.
22871 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22873 * tree.c (build_decl_stat): Remove _stat from name.
22874 * tree.h (build_decl): Remove macro.
22876 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22878 * gimple.c (gimple_build_with_ops_stat): Adjust.
22879 (gimple_alloc_stat): Remove _stat from name.
22880 * gimple.h (gimple_alloc): Remove macro.
22882 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22884 * tree.c (make_tree_vec_stat): Remove _stat from name.
22885 (grow_tree_vec_stat): Likewise.
22886 * tree.h (make_tree_vec_stat): Adjust prototype.
22887 (grow_tree_vec_stat): Likewise.
22888 (make_tree_vec): Remove macro.
22889 (grow_tree_vec): Likewise.
22891 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22893 * fold-const.c (fold_build1_stat_loc): Adjust.
22894 (fold_build2_stat_loc): Likewise.
22895 (fold_build3_stat_loc): Likewise.
22896 * tree.c (build0_stat): Remove _stat from name.
22897 (build1_stat): Likewise.
22898 (build2_stat): Likewise.
22899 (build3_stat): Likewise.
22900 (build4_stat): Likewise.
22901 (build5_stat): Likewise.
22902 * tree.h (build1_loc): Remove macro, and rename _stat function
22904 (build2_loc): Likewise.
22905 (build3_loc): Likewise.
22906 (build4_loc): Likewise.
22907 (build5_loc): Likewise.
22909 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22911 * tree.c (make_int_cst_stat): Remove _stat from name.
22912 * tree.h (make_int_cst_stat): Adjust prototype.
22913 (make_int_cst): Remove macro.
22915 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22917 * tree.c (make_tre_binfo_stat): Remove _stat from name.
22918 * tree.h (make_tree_binfo_stat): Adjust prototype.
22919 (make_tree_binfo): Remove.
22921 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22923 * tree.c (copy_node_stat): Rename to copy_node.
22924 (build_distinct_type_copy): Adjust.
22925 * tree.h (copy_node_stat): Adjust prototype.
22926 (copy_node): Remove macro.
22928 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22930 * tree.c (make_node_stat): rename to make_node.
22931 (build_tree_list_stat): Adjust.
22932 (build0_stat): Likewise.
22933 (build2_stat): Likewise.
22934 (build3_stat): Likewise.
22935 (build4_stat): Likewise.
22936 (build5_stat): Likewise.
22937 (build_decl_stat): Likewise.
22938 * tree.h (make_node_stat): Adjust prototype.
22939 (make_node): remove macro.
22941 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
22943 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
22944 (PPC_FEATURE2_SCV): Likewise.
22945 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
22947 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22949 * config/aarch64/aarch64.c
22950 (aarch64_internal_mov_immediate): Add new special pattern.
22951 * config/aarch64/aarch64.md (*movdi_aarch64):
22952 Add reg/32bit const mov case.
22954 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22955 Richard Sandiford <richard.sandiford@linaro.org>
22957 * config/aarch64/aarch64.md (mov<mode>): Generalize.
22958 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
22959 Add integer and movi cases.
22960 (movi-split-hf-df-sf split, fp16): New.
22961 (enabled): Added TARGET_FP_F16INST.
22962 * config/aarch64/iterators.md (GPF_HF): New.
22963 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
22965 2017-07-28 Tamar Christina <tamar.christina@arm.com>
22967 * config/aarch64/aarch64.c
22968 (aarch64_simd_container_mode): Add prototype.
22969 (aarch64_expand_mov_immediate): Add HI support.
22970 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
22971 (aarch64_can_const_movi_rtx_p): New.
22972 (aarch64_preferred_reload_class):
22973 Remove restrictions of using FP registers for certain SIMD operations.
22974 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
22975 (aarch64_valid_floating_const): Add integer move validation.
22976 (aarch64_simd_imm_scalar_p): Remove.
22977 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
22978 (aarch64_legitimate_constant_p): Expand list of supported cases.
22979 * config/aarch64/aarch64-protos.h
22980 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
22981 (aarch64_reinterpret_float_as_int): New.
22982 (aarch64_simd_imm_scalar_p): Remove.
22983 * config/aarch64/constraints.md (Uvi): New.
22984 (Dd): Split into Ds and new Dd.
22985 * config/aarch64/aarch64.md (*movsi_aarch64):
22987 (*movdi_aarch64): Add SIMD mov case.
22989 2017-07-28 Bin Cheng <bin.cheng@arm.com>
22991 * tree-predcom.c: (struct chain): Handle store-store chain in which
22992 stores for elimination only store loop invariant values.
22993 (execute_pred_commoning_chain): Ditto.
22994 (prepare_initializers_chain_store_elim): Ditto.
22995 (prepare_finalizers): Ditto.
22996 (is_inv_store_elimination_chain): New function.
22997 (initialize_root_vars_store_elim_1): New function.
22999 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23001 * tree-predcom.c: Revise general description of the pass.
23002 (enum chain_type): New enum type for store elimination.
23003 (struct chain): New field supporting store elimination.
23004 (struct component): Ditto.
23005 (dump_chain): Dump store-stores chain.
23006 (release_chain): Release resources.
23007 (split_data_refs_to_components): Compute and create component
23008 contains only stores for elimination.
23009 (get_chain_last_ref_at): New function.
23010 (make_invariant_chain): Initialization.
23011 (make_rooted_chain): Specify chain type in parameter and record it.
23012 (add_looparound_copies): Skip for store-stores chain.
23013 (determine_roots_comp): Compute type of chain and pass it to
23015 (initialize_root_vars_store_elim_2): New function.
23016 (finalize_eliminated_stores): New function.
23017 (remove_stmt): Handle store for elimination.
23018 (execute_pred_commoning_chain): Execute predictive commoning on
23019 store-store chains.
23020 (determine_unroll_factor): Skip unroll for store-stores chain.
23021 (prepare_initializers_chain_store_elim): New function.
23022 (prepare_initializers_chain): Hanlde store-store chain.
23023 (prepare_finalizers_chain, prepare_finalizers): New function.
23024 (tree_predictive_commoning_loop): Return integer value indicating
23025 if loop is unrolled or lcssa form is corrupted.
23026 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
23028 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23030 * tree-predcom.c (initialize_root): Delete.
23031 (execute_pred_commoning_chain): Initialize root vars and replace
23032 reference of non-combined chain directly, rather than call above
23035 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23037 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
23038 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
23040 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23042 * tree-predcom.c (struct chain): New field init_seq.
23043 (release_chain): Release init_seq.
23044 (prepare_initializers_chain): Record intialization stmts in above
23046 (insert_init_seqs): New function.
23047 (tree_predictive_commoning_loop): Call insert_init_seqs.
23049 2017-07-28 Bin Cheng <bin.cheng@arm.com>
23051 * tree-predcom.c (determine_roots_comp): Skip trivial components.
23053 2017-07-28 Richard Biener <rguenther@suse.de>
23055 * match.pd: Remove superfluous :c.
23056 * genmatch.c (simplify::id): Add member.
23057 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
23059 (current_id): New global.
23060 (dt_node::parent): Move from ...
23061 (dt_operand::parent): ... here. Add for_id member.
23062 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
23063 (decision_tree::find_node): Relax order requirement when
23064 merging DT_TRUE nodes to ones inbetween the current simplify
23065 and the one we try to merge with. Add diagnostic whenever
23066 we need to enforce pattern order by not merging.
23067 (decision_tree::insert): Set current_id.
23068 (decision_tree::print_node): Dump parent node and for_id.
23069 (parser::last_id): Add member.
23070 (parser::push_simplify): Assign unique id.
23071 (parser::parser): Initialize last_id.
23073 2017-07-28 Martin Liska <mliska@suse.cz>
23076 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
23077 gimple_build_debug_bind.
23079 2017-07-28 Richard Biener <rguenther@suse.de>
23081 PR tree-optimization/81502
23082 * match.pd: Add pattern combining BIT_INSERT_EXPR with
23084 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
23086 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
23087 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
23088 for BIT_FIELD_REF args.
23089 * fold-const.c (make_bit_field_ref): Likewise.
23090 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
23092 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23095 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
23096 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
23097 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
23098 Or it into SANITIZER_UNDEFINED.
23099 * ubsan.c: Include gimple-fold.h and varasm.h.
23100 (ubsan_expand_ptr_ifn): New function.
23101 (instrument_pointer_overflow): New function.
23102 (maybe_instrument_pointer_overflow): New function.
23103 (instrument_object_size): Formatting fix.
23104 (pass_ubsan::execute): Call instrument_pointer_overflow
23105 and maybe_instrument_pointer_overflow.
23106 * internal-fn.c (expand_UBSAN_PTR): New function.
23107 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
23108 * sanitizer.def (__ubsan_handle_pointer_overflow,
23109 __ubsan_handle_pointer_overflow_abort): New builtins.
23110 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
23111 * internal-fn.def (UBSAN_PTR): New internal function.
23112 * opts.c (sanitizer_opts): Add pointer-overflow.
23113 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
23114 * fold-const.c (build_range_check): Compute pointer range check in
23115 integral type if pointer arithmetics would be needed. Formatting
23118 2017-07-28 Martin Liska <mliska@suse.cz>
23121 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
23122 parameters that are of a variable-length.
23124 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
23126 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
23128 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
23129 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
23130 (CRT_CALL_STATIC_FUNCTION): Likewise.
23131 (ASM_DEFAULT_SPEC): New define.
23132 (ASM_SPEC32): Likewise.
23133 (ASM_SPEC64): Likewise.
23134 (ASM_SPEC_COMMON): Likewise.
23135 (ASM_SPEC): Likewise.
23136 (INVALID_64BIT): Likewise.
23137 (LINK_OS_DEFAULT_SPEC): Likewise.
23138 (LINK_OS_SPEC32): Likewise.
23139 (LINK_OS_SPEC64): Likewise.
23140 (POWERPC_LINUX): Likewise.
23141 (PTRDIFF_TYPE): Likewise.
23142 (RESTORE_FP_PREFIX): Likewise.
23143 (RESTORE_FP_SUFFIX): Likewise.
23144 (SAVE_FP_PREFIX): Likewise.
23145 (SAVE_FP_SUFFIX): Likewise.
23146 (SIZE_TYPE): Likewise.
23147 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
23148 (TARGET_64BIT): Likewise.
23149 (TARGET_64BIT): Likewise.
23150 (TARGET_AIX): Likewise.
23151 (WCHAR_TYPE_SIZE): Likewise.
23152 (WCHAR_TYPE): Undefine.
23153 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
23154 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
23155 (CPP_OS_RTEMS_SPEC): Delete.
23156 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
23157 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
23159 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
23161 2017-07-28 Jakub Jelinek <jakub@redhat.com>
23163 PR tree-optimization/81578
23164 * tree-parloops.c (build_new_reduction): Bail out if
23165 reduction_code isn't one of the standard OpenMP reductions.
23166 Move the details printing after that decision.
23168 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23170 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
23171 related to reload_in_progress.
23172 (splat_input_operand): Likewise.
23173 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
23175 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
23177 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
23178 (TARGET_INSTANTIATE_DECLS): Likewise.
23179 (legitimate_indexed_address_p): Delete reload_in_progress code.
23180 (rs6000_debug_legitimate_address_p): Likewise.
23181 (rs6000_eliminate_indexed_memrefs): Likewise.
23182 (rs6000_emit_le_vsx_store): Likewise.
23183 (rs6000_emit_move_si_sf_subreg): Likewise.
23184 (rs6000_emit_move): Likewise.
23185 (register_to_reg_type): Likewise.
23186 (rs6000_pre_atomic_barrier): Likewise.
23187 (rs6000_machopic_legitimize_pic_address): Likewise.
23188 (rs6000_allocate_stack_temp): Likewise.
23189 (rs6000_address_for_fpconvert): Likewise.
23190 (rs6000_address_for_altivec): Likewise.
23191 (rs6000_secondary_memory_needed_rtx): Delete function.
23192 (rs6000_check_sdmode): Likewise.
23193 (rs6000_alloc_sdmode_stack_slot): Likewise.
23194 (rs6000_instantiate_decls): Likewise.
23195 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
23196 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
23197 Delete reload_in_progress.
23198 (*vec_reload_and_plus_<mptrsize>): Likewise.
23199 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
23200 (vsx_div_v2di): Likewise.
23201 (vsx_udiv_v2di): Likewise.
23203 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23205 * config/rs6000/rs6000.opt (mlra): Replace with stub.
23206 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
23207 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
23208 (rs6000_debug_reg_global): Delete print of LRA status.
23209 (rs6000_option_override_internal): Delete dead LRA related code.
23210 (rs6000_lra_p): Delete function.
23211 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
23213 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
23215 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
23216 * config/riscv/rtems.h: New file.
23218 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23219 Sudakshina Das <sudi.das@arm.com>
23221 * config/aarch64/aarch64.md
23222 (define_split for and<mode>3nr_compare): Move
23223 non aarch64_logical_operand to a register.
23224 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
23225 register immediate operand to a register.
23226 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
23228 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
23230 PR middle-end/81564
23231 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
23233 2017-07-27 Richard Biener <rguenther@suse.de>
23235 PR tree-optimization/81573
23236 PR tree-optimization/81494
23237 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
23238 multi defuse cycle case.
23240 2017-07-27 Richard Biener <rguenther@suse.de>
23242 PR tree-optimization/81571
23243 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
23246 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
23248 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
23249 earlier and only if MASK_FPU is set. Adjust formatting.
23251 2017-07-27 Martin Liska <mliska@suse.cz>
23253 * opt-functions.awk: Add validation of value of Init.
23254 * optc-gen.awk: Pass new argument.
23256 2017-07-27 Martin Liska <mliska@suse.cz>
23258 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
23259 Fix wrong condition.
23261 2017-07-27 Martin Liska <mliska@suse.cz>
23263 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
23264 BBs and edges seen by autoFDO.
23266 2017-07-27 Richard Biener <rguenther@suse.de>
23268 PR tree-optimization/81502
23269 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
23270 with incompatible but same sized type.
23271 (execute_update_addresses_taken): Likewise.
23273 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
23275 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
23276 flag_tree_loop_vectorize rather than flag_tree_vectorize.
23278 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23281 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
23282 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
23283 Change s_operand to memory_operand.
23285 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
23287 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
23288 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
23289 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
23290 Emit instructions rather than returning an expression. Handle TFmode
23291 and KFmode by casting to TImode.
23292 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
23293 (rs6000_emit_le_vsx_store): Likewise.
23294 * config/rs6000/vsx.md (VSX_TI): New iterator.
23295 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
23296 (*vsx_le_undo_permute_<mode>): Likewise.
23297 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
23298 emit the split sequence.
23299 (*vsx_le_perm_store_<mode>): Likewise.
23301 2017-07-27 Jakub Jelinek <jakub@redhat.com>
23303 PR tree-optimization/81555
23304 PR tree-optimization/81556
23305 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
23306 if true, force CHANGED for the recursive invocation.
23307 (reassociate_bb): Remember original length of ops array, pass
23308 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
23310 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
23311 attributes for noipa attribute. For naked attribute use
23312 lookup_attribute first before lookup_attribute_spec.
23313 * final.c (rest_of_handle_final): Disable IPA RA for functions with
23315 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
23316 for functions with noipa attribute.
23317 (cgraph_externally_visible_p): Return true for functions with noipa
23319 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
23320 for functions with noipa attribute.
23321 * doc/extend.texi: Document noipa function attribute.
23322 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
23323 also for functions with noipa attribute.
23324 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
23326 2017-07-26 Andrew Pinski <apinski@cavium.com>
23328 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
23329 vec_unalign_load_cost and vec_unalign_store_cost.
23331 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
23333 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
23334 -mvsx-small-integer option.
23335 (ISA_3_0_MASKS_IEEE): Likewise.
23336 (OTHER_VSX_VECTOR_MASKS): Likewise.
23337 (POWERPC_MASKS): Likewise.
23338 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
23339 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
23340 code, only testing for DImode being allowed in non-VSX floating
23342 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
23343 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
23345 (rs6000_option_override_internal): Delete -mvsx-small-integer.
23346 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
23347 TARGET_P8_VECTOR test.
23348 (rs6000_secondary_reload_simple_move): Likewise.
23349 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
23350 since TARGET_P9_VECTOR was already tested.
23351 (rs6000_opt_masks): Remove -mvsx-small-integer.
23352 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
23353 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23355 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
23356 test for TARGET_VEXTRACTUB was used, and that uses
23358 (p9 extract splitter): Likewise.
23359 (vsx_extract_<mode>_di_p9): Likewise.
23360 (vsx_extract_<mode>_store_p9): Likewise.
23361 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
23362 for TARGET_P9_VECTOR was used. Delete code that is now dead with
23363 the elimination of TARGET_VSX_SMALL_INTEGER.
23364 (vsx_extract_<mode>_p8): Likewise.
23365 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
23366 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
23367 (vsx_set_<mode>_p9): Likewise.
23368 (vsx_set_v4sf_p9): Likewise.
23369 (vsx_set_v4sf_p9_zero): Likewise.
23370 (vsx_insert_extract_v4sf_p9): Likewise.
23371 (vsx_insert_extract_v4sf_p9_2): Likewise.
23372 * config/rs6000/rs6000.md (sign extend splitter): Change
23373 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
23374 (floatsi<mode>2_lfiwax_mem): Likewise.
23375 (floatunssi<mode>2_lfiwzx_mem): Likewise.
23376 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
23377 since a test for TARGET_P9_VECTOR was used.
23378 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23379 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
23380 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23381 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23382 TARGET_P8_VECTOR test.
23383 (fix_trunc<mode>si2_stfiwx): Likewise.
23384 (fix_trunc<mode>si2_internal): Likewise.
23385 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
23386 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23388 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23389 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
23390 TARGET_P8_VECTOR test.
23391 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23392 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
23393 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
23395 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
23396 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
23397 since a test for TARGET_P9_VECTOR was used.
23398 (splitter for loading small constants): Likewise.
23400 2017-07-26 Andrew Pinski <apinski@cavium.com>
23402 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
23405 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
23408 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
23409 (fp_valid_at): Likewise.
23411 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23413 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
23414 (qdf24xx_addrcost_table): Likewise.
23415 (cortexa57_tunings): Update to use generic_branch_cost.
23416 (cortexa72_tunings): Likewise.
23417 (cortexa73_tunings): Likewise.
23418 (qdf24xx_tunings): Likewise.
23420 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
23422 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
23423 (thunderx2t99_branch_cost): Likewise.
23424 (cortexa35_tunings): Update to use generic_branch_cost.
23425 (cortexa53_tunings): Likewise.
23426 (cortexa57_tunings): Likewise.
23427 (cortexa72_tunings): Likewise.
23428 (cortexa73_tunings): Likewise.
23429 (thunderx2t99_tunings): Likewise.
23431 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23433 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
23434 (sparc_option_override): Honour MASK_FSMULD.
23435 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
23436 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
23437 * config/sparc/sparc.opt (mfsmuld): New option.
23438 * doc/invoke.texi (mfsmuld): Document option.
23440 2017-07-26 Marek Polacek <polacek@redhat.com>
23442 PR middle-end/70992
23443 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
23445 2017-07-26 Richard Biener <rguenther@suse.de>
23447 * gimple-match-head.c (do_valueize): Return OP if valueize
23449 (get_def): New helper to get at the def stmt of a SSA name
23450 if valueize allows.
23451 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
23452 do_valueize to get at the def stmt.
23453 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
23455 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
23457 PR middle-end/46932
23458 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
23460 2017-07-26 Martin Liska <mliska@suse.cz>
23463 * function.c (expand_function_start): Make expansion of
23464 nonlocal_goto_save_area after parm_birth_insn.
23466 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
23468 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
23469 from all CPU target flags enable members.
23471 2017-07-26 Richard Biener <rguenther@suse.de>
23473 * genmatch.c (dt_simplify::gen): Make iterator vars const.
23474 (decision_tree::gen): Make 'type' const.
23475 (write_predicate): Likewise.
23477 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23479 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
23480 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
23481 (rs6000_option_override_internal): Likewise.
23482 (rs6000_expand_vector_set): Likewise.
23483 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
23484 (TARGET_UPPER_REGS_SF): Likewise.
23485 (TARGET_UPPER_REGS_DI): Likewise.
23486 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
23487 (TARGET_DIRECT_MOVE_64BIT): Likewise.
23488 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
23489 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
23490 (Splitters for DI constants in Altivec registers): Likewise.
23491 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
23492 (vsx_set_v4sf_p9): Likewise.
23493 (vsx_set_v4sf_p9_zero): Likewise.
23494 (vsx_insert_extract_v4sf_p9): Likewise.
23495 (vsx_insert_extract_v4sf_p9_2): Likewise.
23497 2017-07-25 Carl Love <cel@us.ibm.com>
23499 * doc/extend.texi: Update the built-in documentation file for the
23500 existing built-in functions
23501 vector signed char vec_cnttz (vector signed char);
23502 vector unsigned char vec_cnttz (vector unsigned char);
23503 vector signed short vec_cnttz (vector signed short);
23504 vector unsigned short vec_cnttz (vector unsigned short);
23505 vector signed int vec_cnttz (vector signed int);
23506 vector unsigned int vec_cnttz (vector unsigned int);
23507 vector signed long long vec_cnttz (vector signed long long);
23508 vector unsigned long long vec_cnttz (vector unsigned long long);
23510 2017-07-25 Andrew Pinski <apinski@cavium.com>
23512 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
23513 accesses where the use is for the first operand of a BIT_INSERT.
23515 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23518 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23519 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23521 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
23523 * config/i386/gstabs.h: Delete.
23524 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
23526 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
23528 * config/i386/i386.c (ix86_decompose_address): Do not check for
23529 register RTX when looking at index_reg or base_reg.
23530 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
23532 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
23534 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
23535 to update EH info here.
23537 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23539 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
23541 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23543 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
23545 2017-07-25 Torsten Duwe <duwe@suse.de>
23547 * common.opt: Introduce -fpatchable-function-entry
23548 command line option, and its variables function_entry_patch_area_size
23549 and function_entry_patch_area_start.
23550 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
23551 including a two-value parser.
23552 * target.def (print_patchable_function_entry): New target hook.
23553 * targhooks.h (default_print_patchable_function_entry): New function.
23554 * targhooks.c (default_print_patchable_function_entry): Likewise.
23555 * toplev.c (process_options): Switch off IPA-RA if
23556 patchable function entries are being generated.
23557 * varasm.c (assemble_start_function): Look at the
23558 patchable-function-entry command line switch and current
23559 function attributes and maybe generate NOP instructions by
23560 calling the print_patchable_function_entry hook.
23561 * doc/extend.texi: Document patchable_function_entry attribute.
23562 * doc/invoke.texi: Document -fpatchable_function_entry
23563 command line option.
23564 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
23566 * doc/tm.texi: Re-generate.
23568 2017-07-25 Jakub Jelinek <jakub@redhat.com>
23571 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
23572 TARGET_AVX512DQ rather than TARGET_AVX512BW.
23574 2017-07-25 Tamar Christina <tamar.christina@arm.com>
23576 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
23578 2017-07-25 Richard Biener <rguenther@suse.de>
23580 PR tree-optimization/81455
23581 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
23582 not walk in cycles when looking for guards.
23584 2017-07-25 Richard Biener <rguenther@suse.de>
23586 PR tree-optimization/81529
23587 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
23588 when optimizing backedge uses.
23590 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
23592 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
23594 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
23595 to dl_section_ref. On AIX, append an expression to subtract
23596 the size of the section length to dl_section_ref.
23598 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
23600 * configure.ac: If any of the config.* scripts fail, exit 1.
23601 * configure: Regenerate.
23603 2017-07-25 Richard Biener <rguenther@suse.de>
23605 PR middle-end/81546
23606 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
23607 of immediate uses, be more verbose on errors.
23609 2017-07-25 Richard Biener <rguenther@suse.de>
23611 PR tree-optimization/81510
23612 * tree-vect-loop.c (vect_is_simple_reduction): When the
23613 reduction stmt is not inside the loop bail out.
23615 2017-07-25 Richard Biener <rguenther@suse.de>
23617 PR tree-optimization/81303
23618 * tree-vect-loop-manip.c (vect_loop_versioning): Build
23619 profitability check against LOOP_VINFO_NITERSM1.
23621 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
23623 * domwalk.c (cmp_bb_postorder): Simplify.
23624 (sort_bbs_postorder): New function. Use it...
23625 (dom_walker::walk): ...here to optimize common cases.
23627 2017-07-25 Martin Liska <mliska@suse.cz>
23630 * ipa-visibility.c (function_and_variable_visibility): Make the
23631 redirection just on target that supports aliasing.
23632 Fix GNU coding style.
23634 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
23637 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
23639 * config/arm/rtems.h: Likewise.
23640 * config/bfin/rtems.h: Likewise.
23641 * config/i386/rtemself.h: Likewise.
23642 * config/lm32/rtems.h: Likewise.
23643 * config/m32c/rtems.h: Likewise.
23644 * config/m68k/rtemself.h: Likewise.
23645 * config/microblaze/rtems.h: Likewise.
23646 * config/mips/rtems.h: Likewise.
23647 * config/moxie/rtems.h: Likewise.
23648 * config/nios2/rtems.h: Likewise.
23649 * config/powerpcspe/rtems.h: Likewise.
23650 * config/rs6000/rtems.h: Likewise.
23651 * config/rtems.h: Likewise.
23652 * config/sh/rtems.h: Likewise.
23653 * config/sh/rtemself.h: Likewise.
23654 * config/sparc/rtemself.h: Likewise.
23656 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23659 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
23660 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
23661 * tree-ssa-structalias.c (alias_get_name): Same.
23663 2017-07-25 Bin Cheng <bin.cheng@arm.com>
23666 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
23667 instructions if no du chain is found.
23669 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
23671 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
23673 2017-07-25 Richard Biener <rguenther@suse.de>
23675 PR middle-end/81505
23676 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
23679 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
23681 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
23682 upper-regs options.
23683 (ISA_2_7_MASKS_SERVER): Likewise.
23684 (ISA_3_0_MASKS_IEEE): Likewise.
23685 (OTHER_P8_VECTOR_MASKS): Likewise.
23686 (OTHER_VSX_VECTOR_MASKS): Likewise.
23687 (POWERPC_MASKS): Likewise.
23688 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
23689 duplicate list of options.
23690 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
23691 explicit -mupper-regs options.
23692 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
23693 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
23694 alias for -mupper-regs-df.
23695 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
23696 (rs6000_init_hard_regno_mode_ok): Likewise.
23697 (rs6000_option_override_internal): Likewise.
23698 (rs6000_opt_masks): Likewise.
23699 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
23700 options in terms of whether -mvsx or -mpower8-vector was used.
23701 (TARGET_UPPER_REGS_DI): Likewise.
23702 (TARGET_UPPER_REGS_SF): Likewise.
23703 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
23704 -mupper-regs-* options.
23706 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23708 * passes.c (emergency_dump_function): Print some empty lines and a
23709 header before the RTL dump.
23711 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
23713 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
23715 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
23718 * config/aarch64/aarch64.c (aarch64_classify_symbol):
23719 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
23721 2017-07-24 Carl Love <cel@us.ibm.com>
23723 * config/rs6000/rs6000-c.c: Add support for built-in functions
23724 vector float vec_extract_fp32_from_shorth (vector unsigned short);
23725 vector float vec_extract_fp32_from_shortl (vector unsigned short);
23726 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
23727 vec_extract_fp_from_shortl): Add defines for the two builtins.
23728 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
23729 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
23731 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
23732 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
23733 * doc/extend.texi: Update the built-in documentation file for the
23734 new built-in function.
23736 2017-07-24 Jakub Jelinek <jakub@redhat.com>
23739 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
23741 * doc/generic.texi: Likewise.
23742 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
23743 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
23745 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
23747 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
23748 (aarch64_mls_elt_merge<mode>): Likewise.
23750 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
23752 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
23753 having __cxa_atexit.
23755 2017-07-23 Michael Collison <michael.collison@arm.com>
23757 * config/arm/arm.c (arm_option_override): Deprecate
23758 use of -mstructure-size-boundary.
23759 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
23760 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
23762 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23765 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23766 Reduce cost estimate for direct moves.
23768 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
23771 * config/i386/i386.c (ix86_option_override_internal): Disable
23772 BMI, BMI2 and TBM instructions for -m16.
23774 2017-07-21 Carl Love <cel@us.ibm.com>
23776 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23777 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
23778 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
23779 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
23780 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
23781 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
23782 VMULOSW): New enum "unspec" values.
23783 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
23784 altivec_vmulosw): New patterns.
23785 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
23786 VMULOSW): Add definitions.
23788 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
23790 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
23791 (qdf24xx): Likewise.
23792 * config/aarch64/aarch64-options-extensions.def (rdma); New.
23793 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
23794 (AARCH64_FL_V8_1): Renumber.
23795 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
23796 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
23797 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
23798 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
23799 rdma to feature modifiers list.
23801 2017-07-21 Yury Gribov <tetra2005@gmail.com>
23803 PR middle-end/56727
23804 * ipa-visibility (function_and_variable_visibility): Convert
23805 recursive PLT call to direct call if appropriate.
23807 2017-07-21 Andrew Pinski <apinski@cavium.com>
23809 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
23810 operand 1 to see if the types precision matches.
23811 * fold-const.c (operand_equal_p): Likewise.
23813 2017-07-21 Richard Biener <rguenther@suse.de>
23815 PR tree-optimization/81303
23816 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
23817 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
23818 (vect_peeling_hash_get_lowest_cost): Adjust.
23819 (vect_enhance_data_refs_alignment): Likewise. Use
23820 vect_get_peeling_costs_all_drs to compute the penalty for no
23821 peeling to match up costs.
23823 2017-07-21 Richard Biener <rguenther@suse.de>
23825 PR tree-optimization/81500
23826 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
23827 we didn't identify a reduction path.
23829 2017-07-21 Tom de Vries <tom@codesourcery.com>
23830 Cesar Philippidis <cesar@codesourcery.com>
23832 PR gcov-profile/81442
23833 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
23836 2017-07-21 Tom de Vries <tom@codesourcery.com>
23839 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
23841 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
23842 nvptx_override_options_after_change.
23844 2017-07-21 Ulrich Drepper <drepper@redhat.com>
23846 * dwarf2out.c (output_file_names): Avoid double testing for
23847 dwarf_version >= 5.
23849 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
23851 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
23853 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23855 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
23857 (try_crossjump_to_edge): Do not punt on partitioned functions.
23859 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
23861 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
23862 Put all BBs reachable only via paths crossing cold region to cold
23864 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
23866 2016-07-21 Richard Biener <rguenther@suse.de>
23868 PR tree-optimization/81303
23869 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
23870 into account prologue and epilogue iterations when raising
23871 min_profitable_iters to sth at least covering one vector iteration.
23873 2017-07-21 Tamar Christina <tamar.christina@arm.com>
23875 * config/arm/arm.c (arm_test_cpu_arch_dat):
23878 2017-07-20 Nathan Sidwell <nathan@acm.org>
23880 Remove TYPE_METHODS.
23881 * tree.h (TYPE_METHODS): Delete.
23882 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
23883 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
23884 (dbxout_type_methods): Scan TYPE_FIELDS.
23885 (dbxout_type): Don't check TYPE_METHODS here.
23886 * function.c (use_register_for_decl): Always ignore register for
23887 class types when not optimizing.
23888 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
23889 * tree.c (free_lang_data_in_type): Stitch out member functions and
23890 templates from TYPE_FIELDS.
23891 (build_distinct_type_copy, verify_type_variant,
23892 verify_type): Member fns are on TYPE_FIELDS.
23893 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
23894 * tree-pretty-print.c (dump_generic_node): Likewise.
23896 2017-07-20 Jakub Jelinek <jakub@redhat.com>
23899 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
23900 V2TImode and V4TImode.
23901 (ix86_expand_vector_extract): Likewise.
23902 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
23903 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
23904 (ssescalarmode): Handle V4TImode and V2TImode.
23905 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
23906 (*vec_extractv2ti, *vec_extractv4ti): New insns.
23907 (VEXTRACTI128_MODE): New mode iterator.
23908 (splitter for *vec_extractv?ti first element): New.
23909 (VEC_INIT_MODE): New mode iterator.
23910 (vec_init<mode>): Consolidate 3 expanders into one using
23911 VEC_INIT_MODE mode iterator.
23913 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23915 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
23916 non_spilled_static_chain_regno_p.
23918 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
23920 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
23922 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
23924 * bb-reorder.c (connect_traces): Allow copying of blocks within
23927 2017-07-20 Richard Biener <rguenther@suse.de>
23929 * gimple.h (gimple_phi_result): Add gphi * overload.
23930 (gimple_phi_result_ptr): Likewise.
23931 (gimple_phi_arg): Likewise. Adjust index assert to only
23932 allow actual argument accesses rather than all slots available
23934 (gimple_phi_arg_def): Add gphi * overload.
23935 * tree-phinodes.c (make_phi_node): Initialize only actual
23937 (resize_phi_node): Clear memory not covered by old node,
23938 do not initialize excess argument slots.
23939 (reserve_phi_args_for_new_edge): Initialize new argument slot
23942 2017-07-20 Bin Cheng <bin.cheng@arm.com>
23944 PR tree-optimization/81388
23946 2016-07-21 Bin Cheng <bin.cheng@arm.com>
23948 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
23949 by removing computation of may_be_zero.
23951 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
23952 Tom de Vries <tom@codesourcery.com>
23954 PR middle-end/81030
23955 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
23956 when gimple level profile disagrees with what RTL expander did.
23958 2017-07-20 Richard Biener <rguenther@suse.de>
23960 PR tree-optimization/61171
23961 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
23962 (vect_analyze_stmt): Add slp instance parameter.
23963 (vectorizable_reduction): Likewise.
23964 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
23965 (vect_is_simple_reduction): Deal with chains not detected
23966 as SLP reduction chain, specifically not properly associated
23967 chains containing a mix of plus/minus.
23968 (get_reduction_op): Remove.
23969 (get_initial_defs_for_reduction): Simplify, pass in whether
23970 this is a reduction chain, pass in the SLP node for the PHIs.
23971 (vect_create_epilog_for_reduction): Get the SLP instance as
23973 (vectorizable_reduction): Get the SLP instance as arg.
23974 During analysis remember the SLP node with the PHIs in the
23975 instance. Simplify getting at the vectorized reduction PHIs.
23976 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
23977 through SLP instance.
23978 (vect_slp_analyze_operations): Likewise.
23979 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
23980 (vect_transform_stmt): Likewise.
23982 2017-07-20 Tom de Vries <tom@codesourcery.com>
23984 PR tree-optimization/81489
23985 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
23986 read of phi arg location to before loop that modifies phi.
23988 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
23990 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
23993 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
23995 PR middle-end/81331
23996 * except.c (execute): Fix ordering issue.
23998 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
24000 PR rtl-optimization/81423
24001 * combine.c (make_compound_operation_int): Don't try to optimize
24002 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
24004 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
24006 PR rtl-optimization/81423
24007 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
24008 with a constant that is -1 in the truncated to mode.
24010 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
24012 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
24013 (determine_unlikely_bbs): ... here.
24014 * predict.h (propagate_unlikely_bbs_forward): Declare.
24015 * cfgexpand.c (pass_expand::execute): Use it.
24016 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
24018 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
24019 propagate_unlikely_bbs_forward.
24021 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
24023 PR middle-end/81331
24024 * except.c (maybe_add_nop_after_section_switch): New function.
24027 2017-07-19 Tom de Vries <tom@codesourcery.com>
24029 * gimple.h (gimple_phi_set_arg): Make assert more strict.
24031 2017-07-19 Tom de Vries <tom@codesourcery.com>
24033 * gimple.h (gimple_phi_arg): Make assert more strict.
24035 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
24037 * config.gcc (powerpc*-*-*): Add mmintrin.h.
24038 * config/rs6000/mmintrin.h: New file.
24039 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
24041 2017-07-19 Jakub Jelinek <jakub@redhat.com>
24043 PR tree-optimization/81346
24044 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
24046 2017-07-19 Tom de Vries <tom@codesourcery.com>
24048 * config/nvptx/nvptx.md (VECIM): Add V2DI.
24050 2017-07-19 Tom de Vries <tom@codesourcery.com>
24052 * config/nvptx/nvptx-modes.def: Add V2DImode.
24053 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
24054 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
24055 (nvptx_output_mov_insn): Handle lack of mov.b128.
24056 (nvptx_print_operand): Handle 'H' and 'L' codes.
24057 (nvptx_vector_mode_supported): Allow V2DImode.
24058 (nvptx_preferred_simd_mode): New function.
24059 (nvptx_data_alignment): New function.
24060 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
24061 nvptx_preferred_simd_mode.
24062 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
24064 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
24066 2017-07-19 Tom de Vries <tom@codesourcery.com>
24068 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
24069 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
24070 (nvptx_vector_mode_supported): New function. Allow V2SImode.
24071 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
24072 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
24073 (mov<VECIM>_insn): New define_insn.
24074 (define_expand "mov<VECIM>): New define_expand.
24076 2017-07-19 Tom de Vries <tom@codesourcery.com>
24078 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
24080 2017-07-19 Jakub Jelinek <jakub@redhat.com>
24082 PR tree-optimization/81346
24083 * fold-const.h (fold_div_compare, range_check_type): Declare.
24084 * fold-const.c (range_check_type): New function.
24085 (build_range_check): Use range_check_type.
24086 (fold_div_compare): No longer static, rewritten into
24087 a match.pd helper function.
24088 (fold_comparison): Don't call fold_div_compare here.
24089 * match.pd (X / C1 op C2): New optimization using fold_div_compare
24090 as helper function.
24092 2017-07-19 Nathan Sidwell <nathan@acm.org>
24094 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
24095 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
24096 * tree.c (find_decls_types_r, verify_type): Use
24097 TYPE_{MIN,MAX}_VALUE_RAW.
24098 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
24099 (hash_tree): Likewise.
24100 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
24102 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
24105 2017-07-18 Tom de Vries <tom@codesourcery.com>
24107 PR middle-end/81464
24108 * omp-expand.c (expand_omp_for_static_chunk): Handle
24109 equal-argument loop exit phi.
24111 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
24114 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
24115 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
24116 operand 2 predicate.
24117 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
24118 operand 2 predicate.
24119 (ror,rol -> rorx splitters): Use const_int_operand as
24120 operand 2 predicate.
24122 2017-06-18 Richard Biener <rguenther@suse.de>
24124 PR tree-optimization/81410
24125 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
24126 the gap in the ! slp_perm SLP case after each group.
24128 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24130 PR middle-end/81463
24131 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
24134 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
24136 PR middle-end/81462
24137 * predict.c (set_even_probabilities): Cleanup; do not affect
24138 probabilities that are already known.
24139 (combine_predictions_for_bb): Call even when count is set.
24141 2017-07-18 Nathan Sidwell <nathan@acm.org>
24143 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
24146 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24149 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
24150 optimization for loop niter analysis.
24152 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
24155 * config/avr/avr.c (avr_optimize_casesi): Don't use
24156 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
24158 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
24160 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
24161 body_cost_vec from _vect_peel_extended_info.
24162 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
24163 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
24166 2017-07-18 Bin Cheng <bin.cheng@arm.com>
24168 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
24170 2017-07-18 Richard Biener <rguenther@suse.de>
24172 PR tree-optimization/80620
24173 PR tree-optimization/81403
24174 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
24175 info when re-using a VN table entry.
24177 2017-07-18 Richard Biener <rguenther@suse.de>
24179 PR tree-optimization/81418
24180 * tree-vect-loop.c (vectorizable_reduction): Properly compute
24181 vectype_in. Verify that with lane-reducing reduction operations
24182 we have a single def-use cycle.
24184 2017-07-17 Carl Love <cel@us.ibm.com>
24186 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
24188 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24189 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
24190 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
24191 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
24192 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
24193 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
24194 VMULOSW): New enum "unspec" values.
24195 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
24196 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
24197 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
24198 altivec_vmulosw): New patterns.
24199 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
24200 VMULOSW): Add definitions.
24202 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
24204 * config/alpha/alpha.c: Include predict.h.
24206 2017-07-17 Yury Gribov <tetra2005@gmail.com>
24208 * tree-vrp.c (compare_assert_loc): Fix comparison function
24209 to return predictable results.
24211 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24213 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
24215 (subdi3): Likewise.
24216 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
24217 * doc/invoke.texi (mexpand-adddi): Update text.
24219 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24221 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
24222 that also clobbers the CC register. The old expand code is moved
24224 (*arc_clzsi2): ... here.
24225 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
24226 the CC register. The old expand code is moved to ...
24227 (arc_ctzsi2): ... here.
24229 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24231 * config/arc/arc.opt (mindexed-loads): Use initial value
24232 TARGET_INDEXED_LOADS_DEFAULT.
24233 (mauto-modify-reg): Use initial value
24234 TARGET_AUTO_MODIFY_REG_DEFAULT.
24235 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24236 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24237 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
24238 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
24240 2017-07-17 Martin Liska <mliska@suse.cz>
24243 * opts.c (finish_options): Do not allow -fgnu-tm
24244 w/ -fsanitize={kernel-,}address. Say sorry.
24246 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24249 * tree-loop-distribution.c (classify_partition): Only assert on
24250 numer of iterations.
24251 (merge_dep_scc_partitions): Delete prameter. Update function call.
24252 (distribute_loop): Remove code handling loop with unknown niters.
24253 (pass_loop_distribution::execute): Skip loop with unknown niters.
24255 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24258 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
24259 function sort_partitions_by_post_order.
24261 2017-07-17 Bin Cheng <bin.cheng@arm.com>
24263 PR tree-optimization/81374
24264 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
24265 the max index of basic blocks, rather than number of basic blocks.
24267 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24269 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
24271 (arc_legitimate_pic_operand_p): Likewise.
24272 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
24274 (arc_needs_pcl_p): Likewise.
24275 (arc_legitimate_pc_offset_p): Likewise.
24276 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
24277 function is also used in constrains.md.
24278 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
24279 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
24280 PLUS. Only return true/false in known cases, otherwise assert.
24281 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
24282 is already called in arc_legitimate_constant_p.
24283 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
24285 (LEGITIMATE_PIC_OPERAND_P): Use
24286 arc_raw_symbolic_reference_mentioned_p function.
24287 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
24292 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
24293 Andrew Burgess <andrew.burgess@embecosm.com>
24295 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
24296 (arc_return_address_register): New function.
24297 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
24298 (arc_handle_fndecl_attribute): Add naked attribute.
24299 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
24300 (TARGET_WARN_FUNC_RETURN): Likewise.
24301 (arc_allocate_stack_slots_for_args): New function.
24302 (arc_warn_func_return): Likewise.
24303 (machine_function): Change type fn_type.
24304 (arc_compute_function_type): Consider new naked function type,
24305 change function return type.
24306 (arc_must_save_register): Adapt to handle new
24307 arc_compute_function_type's return type.
24308 (arc_expand_prologue): Likewise.
24309 (arc_expand_epilogue): Likewise.
24310 (arc_return_address_regs): Delete.
24311 (arc_return_address_register): New function.
24312 (arc_epilogue_uses): Use above function.
24313 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
24314 (arc_function_type): Change encoding, add naked type.
24315 (ARC_INTERRUPT_P): Change to handle the new encoding.
24316 (ARC_FAST_INTERRUPT_P): Likewise.
24317 (ARC_NORMAL_P): Define.
24318 (ARC_NAKED_P): Likewise.
24319 (arc_compute_function_type): Delete prototype.
24320 * config/arc/arc.md (in_ret_delay_slot): Use
24321 arc_return_address_register function.
24322 (simple_return): Likewise.
24323 (p_return_i): Likewise.
24325 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24327 PR tree-optimization/81428
24328 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
24329 can't be built for those types.
24331 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24333 Remove stuff dead since r239246.
24335 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
24336 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
24337 (avr_inform_devices): Remove dead stuff.
24339 2017-07-17 Tamar Christina <tamar.christina@arm.com>
24341 * config/arm/arm_neon.h: Fix softp typo.
24343 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24345 PR tree-optimization/81365
24346 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
24347 aggregate moves onto bb predecessor edges, make sure there are no
24348 loads that could alias the lhs in between the start of bb and the
24351 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
24354 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
24355 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
24356 [LSHIFTRT, outer_code = TRUNCATE]: Same.
24358 2017-07-17 Jakub Jelinek <jakub@redhat.com>
24360 PR tree-optimization/81396
24361 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
24362 (init_symbolic_number): Initialize it to 1.
24363 (perform_symbolic_merge): Add n_ops from both operands into the new
24365 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
24366 without base_addr as useless if they need more than one operation.
24367 (bswap_replace): Handle !bswap case for NULL base_addr.
24369 2017-07-17 Tom de Vries <tom@codesourcery.com>
24372 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
24375 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
24377 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
24378 conditional builtin define __FIX_LEON3FT_B2BST.
24380 2017-07-17 Daniel Cederman <cederman@gaisler.com>
24382 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
24383 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
24386 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24388 PR rtl-optimization/81424
24389 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
24390 to remove potential trapping from operands if -fnon-call-exceptions.
24392 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24394 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
24395 profile_proability for scalling.
24396 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
24398 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24400 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
24402 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24404 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
24405 fixpoint arithmetics.
24407 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24409 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
24410 fixpoint arithmetics.
24412 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24414 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
24415 fixpoint arithmetics.
24417 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
24419 * profile-count.h (profile_probability::from_reg_br_prob_note,
24420 profile_probability::to_reg_br_prob_note): New functions.
24421 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
24422 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
24423 * predict.c (probability_reliable_p): Update.
24424 (edge_probability_reliable_p): Update.
24425 (br_prob_note_reliable_p): Update.
24426 (invert_br_probabilities): Update.
24427 (add_reg_br_prob_note): New function.
24428 (combine_predictions_for_insn): Update.
24429 * asan.c (asan_clear_shadow): Update.
24430 * cfgbuild.c (compute_outgoing_frequencies): Update.
24431 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
24432 (update_br_prob_note): Update.
24433 (rtl_verify_edges): Update.
24434 (purge_dead_edges): Update.
24435 (fixup_reorder_chain): Update.
24436 * emit-rtl.c (try_split): Update.
24437 * ifcvt.c (cond_exec_process_insns): Update.
24438 (cond_exec_process_if_block): Update.
24439 (dead_or_predicable): Update.
24440 * internal-fn.c (expand_addsub_overflow): Update.
24441 (expand_neg_overflow): Update.
24442 (expand_mul_overflow): Update.
24443 * loop-doloop.c (doloop_modify): Update.
24444 * loop-unroll.c (compare_and_jump_seq): Update.
24445 * optabs.c (emit_cmp_and_jump_insn_1): Update.
24446 * predict.h: Update.
24447 * reorg.c (mostly_true_jump): Update.
24449 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
24450 * config/alpha/alpha.c (emit_unlikely_jump): Update.
24451 * config/arc/arc.c: (emit_unlikely_jump): Update.
24452 * config/arm/arm.c: (emit_unlikely_jump): Update.
24453 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
24454 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
24455 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
24456 (ix86_print_operand): Update.
24457 (ix86_split_fp_branch): Update.
24458 (predict_jump): Update.
24459 * config/ia64/ia64.c (ia64_print_operand): Update.
24460 * config/mmix/mmix.c (mmix_print_operand): Update.
24461 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
24462 (rs6000_expand_split_stack_prologue): Update.
24463 * config/rs6000/rs6000.c: Update.
24464 * config/s390/s390.c (s390_expand_vec_strlen): Update.
24465 (s390_expand_vec_movstr): Update.
24466 (s390_expand_cs_tdsi): Update.
24467 (s390_expand_split_stack_prologue): Update.
24468 * config/sh/sh.c (sh_print_operand): Update.
24469 (expand_cbranchsi4): Update.
24470 (expand_cbranchdi4): Update.
24471 * config/sparc/sparc.c (output_v9branch): Update.
24472 * config/spu/spu.c (get_branch_target): Update.
24473 (ea_load_store_inline): Update.
24474 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
24475 * config/tilepro/tilepro.c: Update.
24477 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
24479 * gimplify.c (mostly_copy_tree_r): Revert latest change.
24480 (gimplify_save_expr): Likewise.
24482 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24484 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
24486 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24488 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
24490 * timevar.def (TV_IPA_FNSUMMARY): Define.
24492 2017-07-16 Daniel Cederman <cederman@gaisler.com>
24494 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
24495 to back store errata sensitive sequence from being generated.
24496 (sqrtdf2_fix): Likewise.
24498 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24500 * tree-ssa-threadupdate.c (compute_path_counts,
24501 update_joiner_offpath_counts): Use profile_probability.
24503 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
24506 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24508 * config/arm/arm-c.c (arm_cpu_builtins): Define
24509 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24511 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
24513 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24514 array entries to represent __ieee128 versions of the
24515 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
24516 scalar_extract_sig, and scalar_insert_exp built-in functions.
24517 (altivec_resolve_overloaded_builtin): Add special case handling
24518 for the __builtin_scalar_insert_exp function, as represented by
24519 the P9V_BUILTIN_VEC_VSIEDP constant.
24520 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
24521 exponent support for __ieee128 argument.
24522 (VSESQP): Add scalar extract signature support for __ieee128
24524 (VSTDCNQP): Add scalar test negative support for __ieee128
24526 (VSIEQP): Add scalar insert exponent support for __int128 argument
24527 with __ieee128 result.
24528 (VSIEQPF): Add scalar insert exponent support for __ieee128
24529 argument with __ieee128 result.
24530 (VSTDCQP): Add scalar test data class support for __ieee128
24532 (VSTDCNQP): Add overload support for scalar test negative with
24533 __ieee128 argument.
24534 (VSTDCQP): Add overload support for scalar test data class
24535 __ieee128 argument.
24536 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
24537 UNSPEC_VSX_SXSIGDP.
24538 (UNSPEC_VSX_SIEXPQP): New constant.
24539 (xsxexpqp): New insn for VSX scalar extract exponent quad
24541 (xsxsigqp): New insn for VSX scalar extract significand quad
24543 (xsiexpqpf): New insn for VSX scalar insert exponent quad
24544 precision with floating point argument.
24545 (xststdcqp): New expand for VSX scalar test data class quad
24547 (xststdcnegqp): New expand for VSX scalar test negative quad
24549 (xststdcqp): New insn to match expansions for VSX scalar test data
24550 class quad precision and VSX scalar test negative quad precision.
24551 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
24552 special case operand checking to enforce that second operand of
24553 VSX scalar test data class with quad precision argument is a 7-bit
24555 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24556 prototypes and descriptions of __ieee128 versions of
24557 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
24558 scalar_test_data_class, and scalar_test_neg built-in functions.
24560 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24562 PR tree-optimization/81162
24563 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
24564 replace a negate with an add.
24566 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
24568 * doc/invoke.texi (arm/-mcpu): Document +crypto.
24570 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24572 * config/arm/arm-c.c (arm_cpu_builtins): Define
24573 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
24575 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24577 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
24578 (armv8-r): Set ARM Cortex-R52 as default CPU.
24579 * config/arm/arm-tables.opt: Regenerate.
24580 * config/arm/arm-tune.md: Regenerate.
24581 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
24583 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
24584 extension for -mcpu=cortex-r52.
24586 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
24588 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
24589 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
24590 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
24591 (fp-armv8): Define it as FP_ARMv8 only.
24592 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
24593 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
24595 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
24596 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
24597 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
24598 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
24599 than TARGET_FPU_ARMV8.
24600 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
24601 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
24602 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
24604 * config/arm/neon.md (neon_vrint): Likewise.
24605 (neon_vcvt): Likewise.
24606 (neon_<fmaxmin_op><mode>): Likewise.
24607 (<fmaxmin><mode>3): Likewise.
24608 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
24609 * config/arm/predicates.md (arm_cond_move_operator): Check against
24610 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
24612 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
24614 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
24615 to top of function.
24617 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24619 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
24620 loop in comment with memset.
24622 2017-07-14 Martin Liska <mliska@suse.cz>
24624 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
24625 * dwarf2out.c (is_java): Remove the function.
24626 (output_pubname): Remove usage of the function.
24627 (lower_bound_default): Remove usage of DW_LANG_Java.
24628 (gen_compile_unit_die): Likewise.
24629 * gcc.c: Remove compiler defaults for .java and .zip files.
24630 * gimple-expr.c (remove_suffix): Change as there's no longer
24631 extension than 4-letter one.
24632 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
24633 (gimplify_save_expr): Likewise.
24634 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
24635 as it's possible even for other languages than Java.
24636 * langhooks.h (struct lang_hooks): Remove Java from a comment.
24637 * lto-opts.c (lto_write_options): Remove reference to Java.
24638 * opts.c (strip_off_ending): Update file extension handling.
24639 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
24640 * tree-eh.c (lower_resx): Likewise.
24641 * tree.c (free_lang_data_in_type): Remove dead code.
24642 (find_decls_types_r): Likewise.
24643 (build_common_builtin_nodes): Remove Java from a comment.
24644 (verify_type): Remove dead code.
24645 * varasm.c (assemble_external): Remove Java from a comment.
24647 2017-07-14 Martin Liska <mliska@suse.cz>
24649 * opts.c (finish_options): Add quotes.
24650 (common_handle_option): Likewise.
24652 2017-07-14 Martin Liska <mliska@suse.cz>
24654 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
24655 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
24656 Remove N_SO_PASCAL.
24657 * dwarf2out.c (lower_bound_default): Do not handle
24659 (gen_compile_unit_die): Likewise.
24660 * gcc.c: Remove default extension binding for GNU Pascal.
24661 * stmt.c: Remove Pascal language from a comment.
24662 * xcoffout.c: Likewise.
24664 2017-07-13 David Malcolm <dmalcolm@redhat.com>
24667 * diagnostic-show-locus.c (fixit_cmp): New function.
24668 (layout::layout): Sort m_fixit_hints.
24669 (column_range::column_range): Assert that the values are valid.
24670 (struct char_span): New struct.
24671 (correction::overwrite): New method.
24672 (struct source_line): New struct.
24673 (line_corrections::add_hint): Add assertions. Reimplement memcpy
24674 calls in terms of classes source_line and char_span, and
24675 correction::overwrite.
24676 (selftest::test_overlapped_fixit_printing_2): New function.
24677 (selftest::diagnostic_show_locus_c_tests): Call it.
24679 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
24681 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
24682 early if there is no lhs.
24684 2017-07-13 Martin Liska <mliska@suse.cz>
24686 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
24687 (gen_reference_type_die): Likewise.
24688 * stor-layout.c: Remove Pascal-related comment.
24690 2017-07-13 Martin Liska <mliska@suse.cz>
24692 * opts.c (finish_options): Add quotes to error messages.
24693 (parse_sanitizer_options): Likewise.
24695 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
24697 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
24699 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
24701 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
24703 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24705 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
24707 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
24709 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24712 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
24713 provides the hardware capability bits, define the macro
24714 __BUILTIN_CPU_SUPPORTS__.
24715 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
24716 if GLIBC does not provide the hardware capability bits. Add a
24717 gcc_unreachable call if the built-in cpu function is neither
24718 __builtin_cpu_is nor __builtin_cpu_supports.
24719 (rs6000_get_function_versions_dispatcher): Change the warning
24720 that an old GLIBC is used which does not export the capability
24721 bits to be an error.
24722 * doc/extend.texi (target_clones attribute): Document the
24723 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
24724 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
24725 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
24726 the macros defined by GCC if the newer GLIBC is available.
24728 2017-07-12 Jeff Law <law@redhat.com>
24730 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
24731 remaining includes slightly.
24732 * config/riscv/riscv-builtins.c: Include profile-count.h.
24734 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24737 * config/avr/avr.c (avr_set_current_function): In diagnostic
24738 messages: Quote keywords and (parts of) identifiers.
24739 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
24742 2017-07-12 Carl Love <cel@us.ibm.com>
24744 * config/rs6000/rs6000-c.c: Add support for built-in functions
24745 vector bool char vec_revb (vector bool char);
24746 vector bool short vec_revb (vector short char);
24747 vector bool int vec_revb (vector bool int);
24748 vector bool long long vec_revb (vector bool long long);
24749 * doc/extend.texi: Update the built-in documentation file for the
24750 new built-in functions.
24752 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24754 * config/s390/s390.md: Remove movcc splitter.
24756 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24758 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
24759 load/store on condition.
24761 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
24764 * config/avr/avr.c (avr_encode_section_info)
24765 [progmem && !TREE_READONLY]: Error if progmem object needs
24768 2017-07-11 Michael Collison <michael.collison@arm.com>
24770 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
24773 2017-07-11 Carl Love <cel@us.ibm.com>
24775 * config/rs6000/rs6000-c.c: Add support for builtins
24776 vector unsigned int vec_parity_lsbb (vector signed int);
24777 vector unsigned int vec_parity_lsbb (vector unsigned int);
24778 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
24779 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
24780 vector unsigned long long vec_parity_lsbb (vector signed long long);
24781 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
24782 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
24783 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
24784 * doc/extend.texi: Update the built-in documentation file for the
24785 new built-in functions.
24787 2017-07-11 David Malcolm <dmalcolm@redhat.com>
24789 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
24790 (layout::m_primary_loc): New field.
24791 (layout::layout): Initialize new field. Move location filtering
24792 logic from here to...
24793 (layout::maybe_add_location_range): ...this new method. Add
24794 support for filtering to just the lines already specified by other
24796 (layout::will_show_line_p): New method.
24797 (gcc_rich_location::add_location_if_nearby): New method.
24798 (selftest::test_add_location_if_nearby): New test function.
24799 (selftest::diagnostic_show_locus_c_tests): Call it.
24800 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
24803 2017-07-11 Tom de Vries <tom@codesourcery.com>
24805 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
24806 (bb_first_real_insn): New function.
24807 (nvptx_single): Add extra initialization of broadcasted condition
24810 2017-07-11 Nathan Sidwell <nathan@acm.org>
24812 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
24814 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
24816 * doc/extend.texi (AVR Function Attributes): Remove weblink to
24817 Binutils doc as TEXI will mess them up.
24818 * doc/invoke.texi (AVR Options): Same here.
24820 2017-07-11 Daniel Cederman <cederman@gaisler.com>
24822 * config/sparc/sparc.opt (mfix-ut700): New option.
24823 (mfix-gr712rc): Likewise.
24824 (sparc_fix_b2bst): New variable.
24825 * doc/invoke.texi (SPARC options): Document them.
24826 (ARM options): Fix warnings.
24827 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
24828 instructions to prevent sequences that can trigger the store-store
24829 errata for certain LEON3FT processors.
24830 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
24831 (sparc_option_override): Set sparc_fix_b2bst appropriately.
24832 * config/sparc/sparc.md (fix_b2bst): New attribute.
24833 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
24835 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
24838 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
24840 (*rsqrtsf2_sse): Ditto.
24842 (div<mode>3): Macroize insn from divdf3 and divsf3
24843 using MODEF mode iterator.
24845 2017-07-10 Martin Sebor <msebor@redhat.com>
24847 PR tree-optimization/80397
24848 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
24849 instead of testing for equality to INTEGER_TYPE.
24851 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
24853 * config.gcc: Remove uclibc from arc target spec.
24855 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
24857 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
24859 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
24862 * lto-wrapper.c (remove_option): New function.
24863 (merge_and_complain): Merge PIC/PIE options more realistically.
24865 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24867 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
24871 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
24872 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
24873 * config.in: Regenerate.
24874 * configure: Regenerate.
24875 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
24876 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
24877 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
24878 (TARGET_GASISR_PROLOGUES): ...target mask.
24879 * common/config/avr/avr-common.c
24880 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
24881 Set -mgas-isr-prologues.
24882 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
24883 INSERT_PASS_BEFORE for it.
24884 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
24885 * config/avr/avr.c (avr_option_override)
24886 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
24887 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
24888 (avr_attribute_table) <no_gccisr>: Add new function attribute.
24889 (avr_set_current_function) <is_no_gccisr>: Init machine field.
24890 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
24892 (make_avr_pass_pre_proep): New function.
24893 (emit_push_sfr) <treg>: Add argument to function and use it
24894 instead of TMP_REG.
24895 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
24896 and set machine->gasisr.yes.
24897 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
24898 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
24899 __gcc_isr.n_pushed to .L__stack_usage.
24900 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
24901 (avr_asm_final_postscan_insn): ...this new static function.
24902 * config/avr/avr.h (machine_function)
24903 <is_no_gccisr, use_L__stack_usage>: New fields.
24904 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
24905 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
24906 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
24907 (gasisr, *gasisr): New expander and insn.
24908 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
24909 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
24910 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
24912 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
24914 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
24917 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
24919 Move jump-tables out of .text again.
24922 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
24923 (ASM_OUTPUT_ADDR_VEC): New function.
24924 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
24925 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
24926 INSN_ADDRESSes as asm comment.
24927 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
24928 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
24929 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
24930 * config/avr/avr.md (*tablejump): Adjust comment.
24931 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
24932 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
24934 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
24935 (avr_output_addr_vec): New proto.
24936 (avr_log_t) <insn_addresses>: New field.
24938 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
24941 * config/i386/i386.c (ix86_function_arg_advance): Set
24942 outgoing_args_on_stack to true if there are outgoing arguments
24944 (ix86_function_arg): Likewise.
24945 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
24946 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
24947 * config/i386/i386.h (machine_function): Add
24948 outgoing_args_on_stack.
24950 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
24952 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
24953 supporting pthreds.
24954 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
24956 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
24958 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
24959 (REAL_H): Remove $(MACHMODE_H).
24960 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
24962 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
24963 $(MACHMODE_H) and double-int.h.
24964 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
24966 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
24969 2017-07-07 Andrew Pinski <apinski@cavium.com>
24971 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
24972 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
24974 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
24976 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
24977 Add warning if GCC was not configured to link against a GLIBC that
24978 exports the hardware capability bits.
24979 (make_resolver_func): Make resolver function private and not a
24980 COMDAT function. Create the name with clone_function_name instead
24981 of make_unique_name.
24984 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
24985 correct operand in doing the split.
24987 2017-07-07 Carl Love <cel@us.ibm.com>
24989 * config/rs6000/rs6000-c: Add support for built-in function
24990 vector unsigned short vec_pack_to_short_fp32 (vector float,
24992 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
24993 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
24994 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
24995 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
24996 (convert_4f32_8i16): Add define_expand.
24997 * doc/extend.texi: Update the built-in documentation file for the
24998 new built-in function.
25000 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25002 * config/sparc/m8.md: New file.
25003 * config/sparc/sparc.md: Include m8.md.
25005 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25007 * config/sparc/sparc.opt: New option -mvis4b.
25008 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
25009 (sparc_option_override): Handle VIS4B.
25010 (enum sparc_builtins): Define
25011 SPARC_BUILTIN_DICTUNPACK{8,16,32},
25012 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
25013 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
25014 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
25015 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
25016 (check_constant_argument): New function.
25017 (sparc_vis_init_builtins): Define builtins
25018 __builtin_vis_dictunpack{8,16,32},
25019 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
25020 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
25021 __builtin_vis_fpcmpde{8,16,32}shl and
25022 __builtin_vis_fpcmpur{8,16,32}shl.
25023 (sparc_expand_builtin): Check that the constant operands to
25024 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
25025 constant and in range.
25026 * config/sparc/sparc-c.c (sparc_target_macros): Handle
25028 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
25029 (SPARC_IMM5_P): Likewise.
25030 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
25031 (enabled): Handle vis4b.
25032 (UNSPEC_DICTUNPACK): New unspec.
25033 (UNSPEC_FPCMPSHL): Likewise.
25034 (UNSPEC_FPUCMPSHL): Likewise.
25035 (UNSPEC_FPCMPDESHL): Likewise.
25036 (UNSPEC_FPCMPURSHL): Likewise.
25037 (cpu_feature): New CPU feature `vis4b'.
25038 (dictunpack{8,16,32}): New insns.
25039 (FPCSMODE): New mode iterator.
25040 (fpcscond): New code iterator.
25041 (fpcsucond): Likewise.
25042 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
25043 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
25044 (fpcmpde{8,16,32}{si,di}shl): Likewise.
25045 (fpcmpur{8,16,32}{si,di}shl): Likewise.
25046 * config/sparc/constraints.md: Define constraints `q' for unsigned
25047 2-bit integer constants and `t' for unsigned 5-bit integer
25049 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
25051 (imm5_operand_dictunpack16): Likewise.
25052 (imm5_operand_dictunpack32): Likewise.
25053 (imm2_operand): Likewise.
25054 * doc/invoke.texi (SPARC Options): Document -mvis4b.
25055 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
25056 ditunpack* and fpcmp*shl builtins.
25058 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25060 * config.gcc: Handle m8 in --with-{cpu,tune} options.
25061 * config.in: Add HAVE_AS_SPARC6 define.
25062 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
25064 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
25066 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
25067 (CPP_CPU_SPEC): Handle m8.
25068 (ASM_CPU_SPEC): Likewise.
25069 * config/sparc/sparc-opts.h (enum processor_type): Add
25071 * config/sparc/sparc.c (m8_costs): New struct.
25072 (sparc_option_override): Handle TARGET_CPU_m8.
25073 (sparc32_initialize_trampoline): Likewise.
25074 (sparc64_initialize_trampoline): Likewise.
25075 (sparc_issue_rate): Likewise.
25076 (sparc_register_move_cost): Likewise.
25077 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
25078 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
25079 (ASM_CPU64_DEFAULT_SPEC): Likewise.
25080 (CPP_CPU_SPEC): Handle M8.
25081 (ASM_CPU_SPEC): Likewise.
25082 (AS_M8_FLAG): Define.
25083 * config/sparc/sparc.md: Add m8 to the cpu attribute.
25084 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
25085 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
25087 * configure: Regenerate.
25088 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
25091 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25093 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
25095 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
25096 ("*movdi_insn_sp32"): Do not set v3pipe.
25097 ("*movsi_insn"): Likewise.
25098 ("*movdi_insn_sp64"): Likewise.
25099 ("*movsf_insn"): Likewise.
25100 ("*movdf_insn_sp32"): Likewise.
25101 ("*movdf_insn_sp64"): Likewise.
25102 ("*zero_extendsidi2_insn_sp64"): Likewise.
25103 ("*sign_extendsidi2_insn"): Likewise.
25104 ("*mov<VM32:mode>_insn"): Likewise.
25105 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25106 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25107 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25108 ("<vlop:code><VL:mode>3"): Likewise.
25109 ("*not_<vlop:code><VL:mode>3"): Likewise.
25110 ("*nand<VL:mode>_vis"): Likewise.
25111 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
25112 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
25113 ("one_cmpl<VL:mode>2"): Likewise.
25114 ("faligndata<VM64:mode>_vis"): Likewise.
25115 ("alignaddrsi_vis"): Likewise.
25116 ("alignaddrdi_vis"): Likweise.
25117 ("alignaddrlsi_vis"): Likewise.
25118 ("alignaddrldi_vis"): Likewise.
25119 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25120 ("bmaskdi_vis"): Likewise.
25121 ("bmasksi_vis"): Likewise.
25122 ("bshuffle<VM64:mode>_vis"): Likewise.
25123 ("cmask8<P:mode>_vis"): Likewise.
25124 ("cmask16<P:mode>_vis"): Likewise.
25125 ("cmask32<P:mode>_vis"): Likewise.
25126 ("pdistn<P:mode>_vis"): Likewise.
25127 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25129 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25131 * config/sparc/sparc.md ("subtype"): New insn attribute.
25132 ("*wrgsr_sp64"): Set insn subtype.
25133 ("*rdgsr_sp64"): Likewise.
25134 ("alignaddrsi_vis"): Likewise.
25135 ("alignaddrdi_vis"): Likewise.
25136 ("alignaddrlsi_vis"): Likewise.
25137 ("alignaddrldi_vis"): Likewise.
25138 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
25139 ("fexpand_vis"): Likewise.
25140 ("fpmerge_vis"): Likewise.
25141 ("faligndata<VM64:mode>_vis"): Likewise.
25142 ("bshuffle<VM64:mode>_vis"): Likewise.
25143 ("cmask8<P:mode>_vis"): Likewise.
25144 ("cmask16<P:mode>_vis"): Likewise.
25145 ("cmask32<P:mode>_vis"): Likewise.
25146 ("fchksm16_vis"): Likewise.
25147 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
25148 ("fmean16_vis"): Likewise.
25149 ("fp<plusminus_insn>64_vis"): Likewise.
25150 ("<plusminus_insn>v8qi3"): Likewise.
25151 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
25152 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
25153 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
25154 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
25155 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
25156 ("*movqi_insn"): Likewise.
25157 ("*movhi_insn"): Likewise.
25158 ("*movsi_insn"): Likewise.
25159 ("movsi_pic_gotdata_op"): Likewise.
25160 ("*movdi_insn_sp32"): Likewise.
25161 ("*movdi_insn_sp64"): Likewise.
25162 ("movdi_pic_gotdata_op"): Likewise.
25163 ("*movsf_insn"): Likewise.
25164 ("*movdf_insn_sp32"): Likewise.
25165 ("*movdf_insn_sp64"): Likewise.
25166 ("*zero_extendhisi2_insn"): Likewise.
25167 ("*zero_extendqihi2_insn"): Likewise.
25168 ("*zero_extendqisi2_insn"): Likewise.
25169 ("*zero_extendqidi2_insn"): Likewise.
25170 ("*zero_extendhidi2_insn"): Likewise.
25171 ("*zero_extendsidi2_insn_sp64"): Likewise.
25172 ("ldfsr"): Likewise.
25173 ("prefetch_64"): Likewise.
25174 ("prefetch_32"): Likewise.
25175 ("tie_ld32"): Likewise.
25176 ("tie_ld64"): Likewise.
25177 ("*tldo_ldub_sp32"): Likewise.
25178 ("*tldo_ldub1_sp32"): Likewise.
25179 ("*tldo_ldub2_sp32"): Likewise.
25180 ("*tldo_ldub_sp64"): Likewise.
25181 ("*tldo_ldub1_sp64"): Likewise.
25182 ("*tldo_ldub2_sp64"): Likewise.
25183 ("*tldo_ldub3_sp64"): Likewise.
25184 ("*tldo_lduh_sp32"): Likewise.
25185 ("*tldo_lduh1_sp32"): Likewise.
25186 ("*tldo_lduh_sp64"): Likewise.
25187 ("*tldo_lduh1_sp64"): Likewise.
25188 ("*tldo_lduh2_sp64"): Likewise.
25189 ("*tldo_lduw_sp32"): Likewise.
25190 ("*tldo_lduw_sp64"): Likewise.
25191 ("*tldo_lduw1_sp64"): Likewise.
25192 ("*tldo_ldx_sp64"): Likewise.
25193 ("*mov<VM32:mode>_insn"): Likewise.
25194 ("*mov<VM64:mode>_insn_sp64"): Likewise.
25195 ("*mov<VM64:mode>_insn_sp32"): Likewise.
25197 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25199 * config/sparc/sparc.md ("type"): New insn type viscmp.
25200 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
25202 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
25203 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
25204 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
25205 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
25207 ("n7_vis_logical_11cycle"): Likewise.
25208 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
25209 * config/sparc/niagara2.md ("niag3_vis": Likewise.
25210 * config/sparc/niagara.md ("niag_vis"): Likewise.
25211 * config/sparc/ultra3.md ("us3_fga"): Likewise.
25212 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
25214 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
25216 * config/sparc/sparc.md: New instruction type `bmask'.
25217 (bmaskdi_vis): Use the `bmask' type.
25218 (bmasksi_vis): Likewise.
25219 * config/sparc/ultra3.md (us3_array): Likewise.
25220 * config/sparc/niagara7.md (n7_array): Likewise.
25221 * config/sparc/niagara4.md (n4_array): Likewise.
25222 * config/sparc/niagara2.md (niag2_vis): Likewise.
25223 (niag3_vis): Likewise.
25224 * config/sparc/niagara.md (niag_vis): Likewise.
25226 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25228 * ipa-comdats.c: Remove optimize check from gate.
25229 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
25230 for functions not optimized.
25231 (ipa_fn_summary_read): Skip optimize check.
25232 (ipa_fn_summary_write): Likewise.
25233 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
25235 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
25237 (can_inline_edge_p): Check flag_pcc_struct_return for match.
25238 (check_callers): Give up on caller which is not optimized.
25239 (inline_small_functions): Likewise.
25240 (ipa_inline): Do not give up when not optimizing.
25241 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
25242 away unoptimizes cdtors.
25243 (whole_program_function_and_variable_visibility): Do
25244 ipa_discover_readonly_nonaddressable_vars in LTO mode.
25245 * ipa.c (process_references): Do not check optimize.
25246 (symbol_table::remove_unreachable_nodes): Update optimize check.
25247 (set_writeonly_bit): Update optimize check.
25248 (pass_ipa_cdtor_merge::gate): Do not check optimize.
25249 (pass_ipa_single_use::gate): Remove.
25251 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25253 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
25254 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
25255 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
25256 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
25257 permute_load, permute_store, adjust_extract, adjust_splat,
25258 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
25259 replace_swap_with_copy, dump_swap_insn_table,
25260 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
25261 recombine_lvx_pattern, recombine_stvx_pattern,
25262 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
25263 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
25264 to file rs6000-p8swap.c.
25265 * config/rs6000/rs6000-p8swap.c: New file.
25266 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
25267 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
25268 and rs6000*-*-* targets.
25270 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25272 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
25274 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25276 * lto-wrapper.c (merge_and_complain): Do not merge
25277 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
25278 fsigned_zeros, ftrapping_math, fwrapv.
25279 (append_compiler_options): Do not track these options.
25280 (append_linker_options): Likewie
25282 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25284 * cgraphunit.c (cgraph_node::finalize_function): When
25285 !flag_toplevel_reorde set no_reorder flag.
25286 (varpool_node::finalize_decl): Likewise.
25287 (symbol_table::compile): Drop no toplevel reorder path.
25289 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25291 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
25292 edges; zero probability is not better than uninitialized.
25294 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25296 * asan.h (asan_sanitize_allocas_p): Declare.
25297 * asan.c (asan_sanitize_allocas_p): New function.
25298 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
25299 (handle_builtin_alloca): Likewise.
25300 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
25301 if !asan_sanitize_allocas_p.
25302 * params.def (asan-instrument-allocas): Add new option.
25303 * params.h (ASAN_PROTECT_ALLOCAS): Define.
25304 * opts.c (common_handle_option): Disable allocas sanitization for
25307 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
25309 * asan.c: Include gimple-fold.h.
25310 (get_last_alloca_addr): New function.
25311 (handle_builtin_stackrestore): Likewise.
25312 (handle_builtin_alloca): Likewise.
25313 (asan_emit_allocas_unpoison): Likewise.
25314 (get_mem_refs_of_builtin_call): Add new parameter, remove const
25315 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
25316 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
25317 (instrument_builtin_call): Pass gimple iterator to
25318 get_mem_refs_of_builtin_call.
25319 (last_alloca_addr): New global.
25320 * asan.h (asan_emit_allocas_unpoison): Declare.
25321 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
25322 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
25323 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
25324 if function calls alloca.
25325 * gimple-fold.c (replace_call_with_value): Remove static keyword.
25326 * gimple-fold.h (replace_call_with_value): Declare.
25327 * internal-fn.c: Include asan.h.
25328 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
25329 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
25331 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25333 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
25334 (C_SELFTEST_FLAGS): New.
25335 (CPP_SELFTEST_FLAGS): New.
25336 (SELFTEST_DEPS): New, from deps of s-selftest.
25337 (C_SELFTEST_DEPS): New, from deps of s-selftest.
25338 (CPP_SELFTEST_DEPS): New.
25339 (selftest): Add dependency on s-selftest-c++.
25340 (s-selftest): Rename to...
25341 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
25342 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
25343 than SELFTEST_FLAGS.
25344 (selftest-gdb): Rename to...
25345 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
25347 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
25348 (selftest-valgrind): Rename to...
25349 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
25351 (selftest-valgrind): Reintroduce as an alias for
25352 selftest-c-valgrind.
25353 (s-selftest-c++): New.
25354 (selftest-c++-gdb): New.
25355 (selftest-c++-valgrind): New.
25357 2017-07-06 Olivier Hainque <hainque@adacore.com>
25359 * gcc.c (process_command): When deciding if undefined variables
25360 should be ignored when processing specs, accept "gcc -v" as well.
25362 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
25364 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
25365 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
25367 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
25369 * config/arm/arm-cpus.in (armv8-r): Add new entry.
25370 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
25371 * config/arm/arm-tables.opt: Regenerate.
25372 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
25374 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
25376 2017-07-06 Carl Love <cel@us.ibm.com>
25378 * ChangeLog: Clean up from mid air collision
25380 2017-07-06 Carl Love <cel@us.ibm.com>
25382 * config/rs6000/rs6000-c.c: Add support for built-in functions
25383 vector signed int vec_subc (vector signed int, vector signed int);
25384 vector signed __int128 vec_subc (vector signed __int128,
25385 vector signed __int128);
25386 vector unsigned __int128 vec_subc (vector unsigned __int128,
25387 vector unsigned __int128);
25388 vector signed int vec_sube (vector signed int, vector signed int,
25389 vector signed int);
25390 vector unsigned int vec_sube (vector unsigned int,
25391 vector unsigned int,
25392 vector unsigned int);
25393 vector signed __int128 vec_sube (vector signed __int128,
25394 vector signed __int128,
25395 vector signed__int128);
25396 vector unsigned __int128 vec_sube (vector unsigned __int128,
25397 vector unsigned __int128,
25398 vector unsigned __int128);
25399 vector signed int vec_subec (vector signed int, vector signed int,
25400 vector signed int);
25401 vector unsigned int vec_subec (vector unsigned int,
25402 vector unsigned int,
25403 vector unsigned int);
25404 vector signed __int128 vec_subec (vector signed __int128,
25405 vector signed __int128,
25406 vector signed__int128);
25407 vector unsigned __int128 vec_subec (vector unsigned __int128,
25408 vector unsigned __int128,
25409 vector unsigned __int128);
25410 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
25411 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
25412 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
25413 BU_ALTIVEC_OVERLOAD_X definitions.
25414 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
25415 * doc/extend.texi: Update the built-in documentation file for the new
25416 built-in functions.
25418 2017-07-06 David Malcolm <dmalcolm@redhat.com>
25421 * diagnostic-show-locus.c (layout::layout): Use start and finish
25422 spelling location for the start and finish of each range.
25423 * genmatch.c (linemap_client_expand_location_to_spelling_point):
25424 Add unused aspect param.
25425 * input.c (expand_location_1): Add "aspect" param, and use it
25426 to access the correct part of the location.
25427 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
25429 (expand_location_to_spelling_point): Likewise.
25430 (linemap_client_expand_location_to_spelling_point): Add "aspect"
25431 param, and pass it to expand_location_1.
25433 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
25435 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
25436 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
25437 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
25438 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
25439 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
25440 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
25441 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
25442 _mm_maskz_getmant_ss): New intrinsics.
25443 (__builtin_ia32_getexpss128_mask): Changed to ...
25444 __builtin_ia32_getexpss128_round ... this.
25445 (__builtin_ia32_getexpsd128_mask): Changed to ...
25446 __builtin_ia32_getexpsd128_round ... this.
25447 * config/i386/i386-builtin-types.def
25448 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
25449 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
25450 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
25451 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
25452 __builtin_ia32_getmantss_mask_round): New builtins.
25453 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
25454 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
25455 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
25456 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
25457 * config/i386/sse.md
25458 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
25459 avx512f_sgetexp<mode><mask_scalar_name>
25460 <round_saeonly_scalar_name> ... this.
25461 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
25462 %0, %1, %2<round_saeonly_op3>}): Changed to ...
25463 vgetexp<ssescalarmodesuffix>
25464 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25465 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
25466 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
25467 avx512f_vgetmant<mode><mask_scalar_name>
25468 <round_saeonly_scalar_name> ... this.
25469 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
25470 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
25471 vgetmant<ssescalarmodesuffix>
25472 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
25473 %0<mask_scalar_operand4>, %1, %2
25474 <round_saeonly_scalar_mask_op4>, %3} ... this.
25475 * config/i386/subst.md (mask_scalar_operand4,
25476 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
25477 round_saeonly_scalar_nimm_predicate): New subst attributes.
25479 2017-07-06 Julia Koval <julia.koval@intel.com>
25481 * config/i386/i386.c (ix86_erase_embedded_rounding):
25482 Remove code for old rounding pattern.
25484 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
25486 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
25488 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
25490 * doc/sourcebuild.texi (Test Directives, Variants of
25491 dg-require-support): Add documentation for dg-require-stack-check.
25493 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
25495 * config/i386/subst.md (mask_scalar, round_scalar,
25496 round_saeonly_scalar): New meta-templates.
25497 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
25498 round_scalar_mask_operand3, round_scalar_mask_op3,
25499 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
25500 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
25501 round_saeonly_scalar_constraint,
25502 round_saeonly_scalar_prefix): New subst attribute.
25503 * config/i386/sse.md
25504 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
25505 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
25506 <round_scalar_name> ... this.
25507 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
25508 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
25509 <round_scalar_name> ... this.
25510 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
25511 <sse>_vm<code><mode>3<mask_scalar_name>
25512 <round_saeonly_scalar_name> ... this.
25513 (v<plusminus_mnemonic><ssescalarmodesuffix>
25514 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25515 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25516 v<plusminus_mnemonic><ssescalarmodesuffix>
25517 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25518 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25519 (v<multdiv_mnemonic><ssescalarmodesuffix>
25520 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
25521 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
25522 v<multdiv_mnemonic><ssescalarmodesuffix>
25523 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25524 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
25525 (v<maxmin_float><ssescalarmodesuffix>
25526 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
25527 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
25528 v<maxmin_float><ssescalarmodesuffix>
25529 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
25530 %0<mask_scalar_operand3>, %1, %<iptr>2
25531 <round_saeonly_scalar_mask_op3>} ... this.
25533 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
25535 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
25536 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
25538 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
25539 Alan Hayward <alan.hayward@arm.com>
25540 David Sherwood <david.sherwood@arm.com>
25542 * combine.c (simplify_if_then_else): Remove "enum" before
25544 * compare-elim.c (can_eliminate_compare): Likewise.
25545 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
25547 (aarch64_lookup_simd_builtin_type): Likewise.
25548 (aarch64_simd_builtin_type): Likewise.
25549 (aarch64_init_simd_builtin_types): Likewise.
25550 (aarch64_simd_expand_args): Likewise.
25551 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
25553 (aarch64_reverse_mask): Likewise.
25554 (aarch64_simd_emit_reg_reg_move): Likewise.
25555 (aarch64_gen_adjusted_ldpstp): Likewise.
25556 (aarch64_ccmp_mode_to_code): Likewise.
25557 (aarch64_operands_ok_for_ldpstp): Likewise.
25558 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25559 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
25561 (aarch64_min_divisions_for_recip_mul): Likewise.
25562 (aarch64_reassociation_width): Likewise.
25563 (aarch64_get_condition_code_1): Likewise.
25564 (aarch64_simd_emit_reg_reg_move): Likewise.
25565 (aarch64_simd_attr_length_rglist): Likewise.
25566 (aarch64_reverse_mask): Likewise.
25567 (aarch64_operands_ok_for_ldpstp): Likewise.
25568 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
25569 (aarch64_gen_adjusted_ldpstp): Likewise.
25570 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
25572 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
25573 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
25574 (arm_lookup_simd_builtin_type): Likewise.
25575 (arm_simd_builtin_type): Likewise.
25576 (arm_init_simd_builtin_types): Likewise.
25577 (arm_expand_builtin_args): Likewise.
25578 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
25579 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
25580 (ft32_setup_incoming_varargs): Likewise.
25581 (ft32_function_arg): Likewise.
25582 (ft32_function_arg_advance): Likewise.
25583 (ft32_pass_by_reference): Likewise.
25584 (ft32_arg_partial_bytes): Likewise.
25585 (ft32_valid_pointer_mode): Likewise.
25586 (ft32_addr_space_pointer_mode): Likewise.
25587 (ft32_addr_space_legitimate_address_p): Likewise.
25588 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
25590 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
25591 (ix86_emit_outlined_ms2sysv_restore): Likewise.
25592 (iamcu_alignment): Likewise.
25593 (canonicalize_vector_int_perm): Likewise.
25594 (ix86_noce_conversion_profitable_p): Likewise.
25595 (ix86_mpx_bound_mode): Likewise.
25596 (ix86_operands_ok_for_move_multiple): Likewise.
25597 * config/microblaze/microblaze-protos.h
25598 (microblaze_expand_conditional_branch_reg): Likewise.
25599 * config/microblaze/microblaze.c
25600 (microblaze_expand_conditional_branch_reg): Likewise.
25601 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
25603 (rs6000_reassociation_width): Likewise.
25604 (rs6000_invalid_binary_op): Likewise.
25605 (fusion_p9_p): Likewise.
25606 (emit_fusion_p9_load): Likewise.
25607 (emit_fusion_p9_store): Likewise.
25608 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
25610 (riscv_hard_regno_mode_ok_p): Likewise.
25611 (riscv_address_insns): Likewise.
25612 (riscv_split_symbol): Likewise.
25613 (riscv_legitimize_move): Likewise.
25614 (riscv_function_value): Likewise.
25615 (riscv_hard_regno_nregs): Likewise.
25616 (riscv_expand_builtin): Likewise.
25617 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
25618 (riscv_build_integer): Likewise.
25619 (riscv_split_integer): Likewise.
25620 (riscv_legitimate_constant_p): Likewise.
25621 (riscv_cannot_force_const_mem): Likewise.
25622 (riscv_regno_mode_ok_for_base_p): Likewise.
25623 (riscv_valid_base_register_p): Likewise.
25624 (riscv_valid_offset_p): Likewise.
25625 (riscv_valid_lo_sum_p): Likewise.
25626 (riscv_classify_address): Likewise.
25627 (riscv_legitimate_address_p): Likewise.
25628 (riscv_address_insns): Likewise.
25629 (riscv_load_store_insns): Likewise.
25630 (riscv_force_binary): Likewise.
25631 (riscv_split_symbol): Likewise.
25632 (riscv_force_address): Likewise.
25633 (riscv_legitimize_address): Likewise.
25634 (riscv_move_integer): Likewise.
25635 (riscv_legitimize_const_move): Likewise.
25636 (riscv_legitimize_move): Likewise.
25637 (riscv_address_cost): Likewise.
25638 (riscv_subword): Likewise.
25639 (riscv_output_move): Likewise.
25640 (riscv_canonicalize_int_order_test): Likewise.
25641 (riscv_emit_int_order_test): Likewise.
25642 (riscv_function_arg_boundary): Likewise.
25643 (riscv_pass_mode_in_fpr_p): Likewise.
25644 (riscv_pass_fpr_single): Likewise.
25645 (riscv_pass_fpr_pair): Likewise.
25646 (riscv_get_arg_info): Likewise.
25647 (riscv_function_arg): Likewise.
25648 (riscv_function_arg_advance): Likewise.
25649 (riscv_arg_partial_bytes): Likewise.
25650 (riscv_function_value): Likewise.
25651 (riscv_pass_by_reference): Likewise.
25652 (riscv_setup_incoming_varargs): Likewise.
25653 (riscv_print_operand): Likewise.
25654 (riscv_elf_select_rtx_section): Likewise.
25655 (riscv_save_restore_reg): Likewise.
25656 (riscv_for_each_saved_reg): Likewise.
25657 (riscv_register_move_cost): Likewise.
25658 (riscv_hard_regno_mode_ok_p): Likewise.
25659 (riscv_hard_regno_nregs): Likewise.
25660 (riscv_class_max_nregs): Likewise.
25661 (riscv_memory_move_cost): Likewise.
25662 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
25663 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
25664 (rl78_addr_space_address_mode): Likewise.
25665 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
25667 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
25668 (rs6000_reassociation_width): Likewise.
25669 (rs6000_invalid_binary_op): Likewise.
25670 (fusion_p9_p): Likewise.
25671 (emit_fusion_p9_load): Likewise.
25672 (emit_fusion_p9_store): Likewise.
25673 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
25674 (ok_for_simple_move_operands): Likewise.
25675 (ok_for_simple_move_strict_operands): Likewise.
25676 (ok_for_simple_arith_logic_operands): Likewise.
25677 (visium_legitimize_reload_address): Likewise.
25678 (visium_select_cc_mode): Likewise.
25679 (output_cbranch): Likewise.
25680 (visium_split_double_move): Likewise.
25681 (visium_expand_copysign): Likewise.
25682 (visium_expand_int_cstore): Likewise.
25683 (visium_expand_fp_cstore): Likewise.
25684 * config/visium/visium.c (visium_pass_by_reference): Likewise.
25685 (visium_function_arg): Likewise.
25686 (visium_function_arg_advance): Likewise.
25687 (visium_libcall_value): Likewise.
25688 (visium_setup_incoming_varargs): Likewise.
25689 (visium_legitimate_constant_p): Likewise.
25690 (visium_legitimate_address_p): Likewise.
25691 (visium_legitimize_address): Likewise.
25692 (visium_secondary_reload): Likewise.
25693 (visium_register_move_cost): Likewise.
25694 (visium_memory_move_cost): Likewise.
25695 (prepare_move_operands): Likewise.
25696 (ok_for_simple_move_operands): Likewise.
25697 (ok_for_simple_move_strict_operands): Likewise.
25698 (ok_for_simple_arith_logic_operands): Likewise.
25699 (visium_function_value_1): Likewise.
25700 (rtx_ok_for_offset_p): Likewise.
25701 (visium_legitimize_reload_address): Likewise.
25702 (visium_split_double_move): Likewise.
25703 (visium_expand_copysign): Likewise.
25704 (visium_expand_int_cstore): Likewise.
25705 (visium_expand_fp_cstore): Likewise.
25706 (visium_split_cstore): Likewise.
25707 (visium_select_cc_mode): Likewise.
25708 (visium_split_cbranch): Likewise.
25709 (output_cbranch): Likewise.
25710 (visium_print_operand_address): Likewise.
25711 * expmed.c (flip_storage_order): Likewise.
25712 * expmed.h (emit_cstore): Likewise.
25713 (flip_storage_order): Likewise.
25714 * genrecog.c (validate_pattern): Likewise.
25715 * hsa-gen.c (gen_hsa_addr): Likewise.
25716 * internal-fn.c (expand_arith_overflow): Likewise.
25717 * ira-color.c (allocno_copy_cost_saving): Likewise.
25718 * lra-assigns.c (find_hard_regno_for_1): Likewise.
25719 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
25720 (process_invariant_for_inheritance): Likewise.
25721 * lra-eliminations.c (move_plus_up): Likewise.
25722 * omp-low.c (lower_oacc_reductions): Likewise.
25723 * simplify-rtx.c (simplify_subreg): Likewise.
25724 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
25725 (TARGET_CHKP_BOUND_MODE): Likewise..
25726 * targhooks.c (default_chkp_bound_mode): Likewise.
25727 (default_setup_incoming_vararg_bounds): Likewise.
25728 * targhooks.h (default_chkp_bound_mode): Likewise.
25729 (default_setup_incoming_vararg_bounds): Likewise.
25730 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
25731 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25732 (have_whole_vector_shift): Likewise.
25733 * tree-vect-stmts.c (vectorizable_load): Likewise.
25734 * doc/tm.texi: Regenerate.
25736 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25738 Graceful degrade if Binutils PR21472 is not available.
25741 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
25742 .rodata in flash test fails.
25743 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
25744 * confgure: Regenerate.
25745 * config.in: Regenerate.
25746 * config/avr/avr.c (avr_asm_named_section)
25747 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
25748 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
25749 (avr_asm_init_sections): Same.
25751 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25753 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
25754 (fma<VH:mode>4_intrinsic): Likewise.
25755 (*fmsub<VCVTF:mode>4): Likewise.
25756 (*fmsub<VH:mode>4_intrinsic): Likewise.
25758 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
25761 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
25762 Don't depend on "optimize > 0".
25763 (out_movhi_r_mr, out_movqi_mr_r): Same.
25764 (out_movhi_mr_r, out_movqi_r_mr): Same.
25765 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
25766 io_address_operand on "optimize > 0".
25768 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25770 * tree-loop-distribution.c: Add general explanantion on the pass.
25771 (generate_loops_for_partition): Mark distributed loop.
25772 (pg_add_dependence_edges): New parameter. Handle alias data
25773 dependence specially and record it in the parameter if asked.
25774 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
25775 (init_partition_graph_vertices, add_partition_graph_edge): New.
25776 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
25777 (free_partition_graph_vdata, build_partition_graph): New.
25778 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
25779 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
25780 (data_ref_segment_size, latch_dominated_by_data_ref): New.
25781 (compute_alias_check_pairs, version_loop_by_alias_check): New.
25782 (version_for_distribution_p, finalize_partitions): New.
25783 (distribute_loop): Handle alias data dependence specially. Factor
25784 out loop fusion code as functions and call these functions.
25786 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25788 * tree-loop-distribution.c (classify_partition): New parameter and
25789 better handle reduction statement.
25790 (rdg_build_partitions): Revise comment.
25791 (distribute_loop): Compute statements in all partitions and pass it
25792 to classify_partition.
25794 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25796 * tree-loop-distribution.c (enum partition_type): New.
25797 (struct partition): New field type.
25798 (partition_merge_into): Add parameter. Update partition type.
25799 (data_dep_in_cycle_p, update_type_for_merge): New functions.
25800 (build_rdg_partition_for_vertex): Compute partition type.
25801 (rdg_build_partitions): Dump partition type.
25802 (distribute_loop): Update calls to partition_merge_into.
25804 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25806 * tree-loop-distribution.c (struct ddr_hasher): New.
25807 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
25809 (classify_partition): Call get_data_dependence.
25810 (pg_add_dependence_edges): Ditto.
25811 (distribute_loop): Release data dependence hash table.
25813 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25815 * tree-loop-distribution.c (ref_base_address): Delete.
25816 (similar_memory_accesses): Rename ...
25817 (share_memory_accesses): ... to this. Check if partitions access
25818 the same memory reference.
25819 (distribute_loop): Call share_memory_accesses.
25821 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25823 * tree-loop-distribution.c (struct partition): New field recording
25824 its data reference.
25825 (partition_alloc, partition_free): Init and release data refs.
25826 (partition_merge_into): Merge data refs.
25827 (build_rdg_partition_for_vertex): Collect data refs for partition.
25828 (pg_add_dependence_edges): Change parameters from vector to bitmap.
25830 (distribute_loop): Remve data refs from vertice data of partition
25833 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25835 * tree-loop-distribution.c (params.h): Include header file.
25836 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
25837 (datarefs_vec): New global var.
25838 (create_rdg_vertices): Use datarefs_vec directly.
25839 (free_rdg): Don't free data references.
25840 (build_rdg): Update use. Don't free data references.
25841 (distribute_loop): Compute global variable for data references.
25842 Bail out if there are too many data references.
25844 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25846 * tree-loop-distribution.c (loop_nest): New global var.
25847 (build_rdg): Use loop directly, rather than loop nest.
25848 (pg_add_dependence_edges): Remove loop nest parameter. Use global
25850 (distribute_loop): Compute global variable loop nest. Update use.
25852 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25854 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
25855 (partition_merge_into): New parameter. Dump reason for fusion.
25856 (distribute_loop): Update use of partition_merge_into.
25858 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25860 * tree-loop-distribution.c (bb_top_order_index): New.
25861 (bb_top_order_index_size, bb_top_order_cmp): New.
25862 (stmts_from_loop): Use topological order.
25863 (pass_loop_distribution::execute): Compute and release topological
25864 order for basic blocks.
25866 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25868 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
25871 2017-07-05 Bin Cheng <bin.cheng@arm.com>
25873 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
25874 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
25875 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
25876 * internal-fn.def (LOOP_DIST_ALIAS): New.
25877 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
25878 (fold_loop_internal_call): ... this.
25879 (vect_loop_dist_alias_call): New function.
25880 (set_uid_loop_bbs): Call fold_loop_internal_call.
25881 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
25884 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25887 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
25888 Require dead FLAGS_REG at the beginning of a peephole.
25890 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
25893 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
25894 arguments in the call to __builtin_ia32_sbb_u32.
25895 (_subborrow_u64): Swap _X and _Y arguments in the call to
25896 __builtin_ia32_sbb_u64.
25898 2017-07-04 Jakub Jelinek <jakub@redhat.com>
25901 * tree-vrp.c (compare_assert_loc): Turn into a function template
25902 with stable template parameter. Only test if a->e is NULL,
25903 !a->e == !b->e has been verified already. Use e == NULL or
25904 e != NULL instead of e or ! e tests. If stable is true, don't use
25905 iterative_hash_expr, on the other side allow a or b or both NULL
25906 and sort the NULLs last.
25907 (process_assert_insertions): Sort using compare_assert_loc<false>
25908 instead of compare_assert_loc, later sort using
25909 compare_assert_loc<true> before calling process_assert_insertions_for
25910 in a loop. Use break instead of continue once seen NULL pointer.
25912 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25914 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
25915 Cortex-R7 and Cortex-R8 processors.
25917 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25919 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
25920 uninitialized while src is not.
25922 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
25924 * common/config/arm/arm-common.c: Adjust include path for
25926 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
25927 (arm-cpu.h): Create in build directory. Adjust dependency rules.
25928 (arm-cpu-data.h): Likewise.
25929 (arm-cpu-cdata.h): Likewise.
25930 * config/arm/arm-cpu.h: Delete.
25931 * config/arm/arm-cpu-cdata.h: Delete.
25932 * config/arm/arm-cpu-data.h: Delete.
25934 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
25936 * config/arm/arm-cpus.in (cortex-a55): New.
25937 (cortex-a75): Likewise.
25938 (cortex-a75.cortex-a55): Likewise.
25939 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
25941 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
25942 * config/arm/arm-cpu-cdata.h: Regenerate.
25943 * config/arm/arm-cpu-data.h: Regenerate.
25944 * config/arm/arm-cpu.h: Regenerate.
25945 * config/arm/arm-tables.opt: Regenerate.
25946 * config/arm/arm-tune.md: Regenerate.
25948 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25950 * haifa-sched.c (sched_create_recovery_edges): Update profile.
25952 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
25954 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
25957 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
25959 PR tree-optimization/81292
25960 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
25961 full_string_p, also call adjust_related_strinfos if the adjustment
25962 is simple, otherwise invalidate related strinfos.
25964 2017-07-04 Martin Liska <mliska@suse.cz>
25967 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
25968 newly created variable as DECL_IGNORED_P.
25970 2017-07-04 Martin Liska <mliska@suse.cz>
25973 * ipa-inline.c (inline_small_functions):
25974 Use xstrdup_for_dump.
25976 2017-07-04 Tom de Vries <tom@codesourcery.com>
25978 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
25980 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
25983 * config/darwin.c (darwin_function_switched_text_sections):
25986 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
25988 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
25990 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
25992 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
25994 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
25996 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
25997 min_profitable_iters, and th as inclusive lower bounds.
25998 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
25999 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
26000 for min_profitable_iters and min_profitable_estimate.
26001 (vect_transform_loop): Treat th as an inclusive lower bound.
26002 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
26004 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
26007 * config/darwin.c (darwin_function_switched_text_sections):
26008 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
26009 in two pieces, and suppress the use of buf.
26011 2017-07-03 Nathan Sidwell <nathan@acm.org>
26013 * hash-table.h (hash_table_mod1): Fix indentation.
26015 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26017 PR middle-end/81290
26018 * predict.c (force_edge_cold): Be more careful about propagation
26020 * profile-count.h (profile_probability::guessed,
26021 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
26023 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
26025 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
26027 * doc/invoke.texi (rcpc architecture extension): Document it.
26029 2017-07-03 Richard Biener <rguenther@suse.de>
26031 PR tree-optimization/60510
26032 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
26033 the scalar reduction PHI and use it.
26034 (vectorizable_reduction): Properly guard the single_defuse_cycle
26035 path for non-SLP reduction chains where we cannot use it.
26036 Rework reduc_def/index and vector type deduction. Rework
26037 vector operand gathering during reduction op code-gen.
26038 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
26039 chains dissolve the chain and leave it to non-SLP reduction
26042 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26044 * tree-data-ref.h (dr_alignment): Declare.
26045 * tree-data-ref.c (dr_alignment): New function.
26046 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
26047 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
26049 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
26051 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26053 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
26054 and base_misalignment fields.
26055 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
26056 * tree-data-ref.c: Include builtins.h.
26057 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
26058 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
26059 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
26060 * tree-vect-data-refs.c: Include tree-cfg.h.
26061 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
26062 fields instead of calculating an alignment here.
26063 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
26064 innermost_loop_behavior fields.
26066 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26068 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
26070 (DR_STEP_ALIGNMENT): New macro.
26071 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
26072 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
26073 (create_data_ref): Print it.
26074 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
26075 to tell whether the step preserves vector (mis)alignment.
26076 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
26077 Move the check for an integer step and generalise to all INTEGER_CST.
26078 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
26079 Print the outer step alignment.
26081 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26083 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
26084 with offset_alignment.
26085 (DR_ALIGNED_TO): Delete.
26086 (DR_OFFSET_ALIGNMENT): New macro.
26087 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
26088 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
26089 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
26090 (create_data_ref): Likewise.
26091 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
26092 (vect_analyze_data_refs): Likewise.
26093 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
26094 creating dummy innermost behavior.
26096 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26098 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
26099 with a "innermost_loop_behavior *" and refeence tree.
26100 * tree-data-ref.c (dr_analyze_innermost): Likewise.
26101 (create_data_ref): Update call accordingly.
26102 * tree-predcom.c (find_looparound_phi): Likewise.
26104 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26106 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
26107 fields with dr_wrt_vec_loop.
26108 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
26109 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
26110 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
26111 (vect_dr_behavior): New function.
26112 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26113 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
26114 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
26115 track whether the step preserves the misalignment.
26116 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
26117 Use vect_dr_behavior.
26118 (vect_setup_realignment): Update call accordingly.
26119 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
26120 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
26121 call to vect_create_addr_base_for_vector_ref.
26122 (vect_create_cond_for_align_checks): Likewise.
26123 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
26124 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
26125 (vect_recog_mask_conversion_pattern): Likewise.
26126 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
26127 (new_stmt_vec_info): Remove redundant zeroing.
26129 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
26131 * common/config/arm/arm-common.c (arm_be8_option): New function.
26132 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
26133 (ISA_ARMv6): Add isa_bit_be8.
26134 * config/arm/arm.h (arm_be8_option): Add prototype.
26135 (BE8_SPEC_FUNCTION): New define.
26136 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
26137 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
26138 (mlittle-endian): Similarly.
26139 (mbe8, mbe32): New options.
26140 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
26141 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
26143 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26145 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
26147 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26149 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
26150 (cleanup_tree_cfg_bb): Use it.
26151 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
26153 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
26155 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26158 * loop-doloop.c (add_test): Update profile.
26160 2017-07-03 Martin Liska <mliska@suse.cz>
26163 * sanopt.c (rewrite_usage_of_param): New function.
26164 (sanitize_rewrite_addressable_params): Likewise.
26165 (pass_sanopt::execute): Call rewrite_usage_of_param.
26167 2017-07-03 Richard Biener <rguenther@suse.de>
26169 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
26170 back to using VIEW_CONVERT_EXPR.
26172 2017-07-03 Martin Liska <mliska@suse.cz>
26175 * doc/extend.texi: Document when a resolver function is
26176 generated for target_clones.
26178 2017-07-03 Martin Liska <mliska@suse.cz>
26180 * asan.c (asan_emit_stack_protection): Unpoison just red zones
26181 and shadow memory of auto variables which are subject of
26182 use-after-scope sanitization.
26183 (asan_expand_mark_ifn): Add do set only when is_poison.
26185 2016-07-03 Richard Biener <rguenther@suse.de>
26187 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
26189 (vect_force_simple_reduction): Record reduction def -> phi mapping.
26190 (vectorizable_reduction): Perform reduction PHI creation when
26191 visiting a reduction PHI and adjust and simplify code generation
26192 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
26193 (vect_transform_loop): Visit reduction PHIs.
26194 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
26195 defs into the SLP tree.
26196 (vect_build_slp_tree): Reduction defs terminate the recursion.
26197 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
26199 (vect_get_vec_defs_for_stmt_copy): Export.
26200 (vect_get_vec_defs): Likewise.
26201 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
26203 (vect_get_vec_defs_for_stmt_copy): Declare.
26204 (vect_get_vec_defs): Likewise.
26206 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
26208 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
26209 parameter with a "loop" parameter and use it instead of the
26210 loop containing DR_STMT. Don't check simple_iv when doing
26211 BB analysis. Describe the two analysis modes in the comment.
26213 2017-07-03 Tom de Vries <tom@codesourcery.com>
26215 PR tree-optimization/69468
26216 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
26217 (find_same_succ_bb): Handle ignore_edge_flags.
26219 2017-07-03 Tom de Vries <tom@codesourcery.com>
26221 PR tree-optimization/81192
26222 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
26224 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
26226 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
26228 2017-07-03 Tom de Vries <tom@codesourcery.com>
26230 PR tree-optimization/81192
26231 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
26232 BB_SAME_SUCC (bb) == NULL.
26234 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26236 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
26239 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26241 * dumpfile.c: Include profile-count.h
26242 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
26244 (insert_cond_bb): Update profile.
26245 * tree-cfg.h (insert_cond_bb): Update prototype.
26246 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
26247 * tree-dump.c: Do not include tree-cfg.
26249 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26251 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
26253 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26255 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
26258 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
26260 * tree-complex.c (expand_complex_div_wide): update profile.
26262 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26263 Alan Hayward <alan.hayward@arm.com>
26264 David Sherwood <david.sherwood@arm.com>
26266 * Makefile.in (MACHMODE_H): Remove insn-modes.h
26267 (CORETYPES_H): New define.
26268 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
26269 (insn-modes-inline.h, s-modes-inline-h): New rules.
26270 (generated_files): Add insn-modes-inline.h.
26271 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
26272 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
26273 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
26274 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
26275 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
26276 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
26277 (build/gencodes.o, build/genconditions.o): Likewise.
26278 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
26279 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
26280 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
26281 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
26282 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
26283 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
26284 * coretypes.h: Include everything up to real.h for generators.
26285 Include insn-modes.h first. Include wide-int-print.h after
26286 wide-int.h. Include insn-modes-inline.h and then machmode.h.
26287 * machmode.h: Don't include insn-modes.h here.
26288 * function-tests.c: Remove includes of signop.h, machmode.h,
26289 double-int.h and wide-int.h.
26291 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
26293 * optc-save-gen.awk: Likewise.
26294 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
26295 * godump.c: Remove include of wide-int-print.h.
26296 * pretty-print.h: Likewise.
26297 * wide-int-print.cc: Likewise.
26298 * wide-int.cc: Likewise.
26299 * hash-map-tests.c: Remove include of signop.h.
26300 * hash-set-tests.c: Likewise.
26301 * rtl-tests.c: Likewise.
26302 * mkconfig.sh: Remove include of machmode.h.
26303 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
26305 (emit_insn_modes_inline_h): ...this new function. Emit the code
26306 into an insn-modes-inline.h header file, adding appropriate
26307 include guards and end comments.
26308 (emit_insn_modes_c_header): Remove include of machmode.h.
26309 (emit_min_insn_modes_c_header): Include coretypes.h rather than
26311 (main): Handle -i flag and call emit_insn_modes_inline_h when
26314 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26316 * tree-ssa-strlen.c (strinfo): Rename the length field to
26317 nonzero_chars. Add a full_string_p field.
26318 (compare_nonzero_chars, zero_length_string_p): New functions.
26319 (get_addr_stridx): Add an offset_out parameter.
26320 Use compare_nonzero_chars.
26321 (get_stridx): Update accordingly. Use compare_nonzero_chars.
26322 (new_strinfo): Update after above changes to strinfo.
26323 (set_endptr_and_length): Set full_string_p.
26324 (get_string_length): Update after above changes to strinfo.
26325 (unshare_strinfo): Update call to new_strinfo.
26326 (maybe_invalidate): Likewise.
26327 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
26328 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
26329 as a uhwi instead of an shwi. Update after above changes to
26330 strinfo and new_strinfo.
26331 (zero_length_string): Assert that chainsi contains full strings.
26332 Use zero_length_string_p. Update call to new_strinfo.
26333 (adjust_related_strinfos): Update after above changes to strinfo.
26334 Copy full_string_p from origsi.
26335 (adjust_last_stmt): Use zero_length_string_p.
26336 (handle_builtin_strlen): Update after above changes to strinfo and
26337 new_strinfo. Install the lhs as the string length if the previous
26338 entry didn't describe a full string.
26339 (handle_builtin_strchr): Update after above changes to strinfo
26341 (handle_builtin_strcpy): Likewise.
26342 (handle_builtin_strcat): Likewise.
26343 (handle_builtin_malloc): Likewise.
26344 (handle_pointer_plus): Likewise.
26345 (handle_builtin_memcpy): Likewise. Track nonzero characters
26346 that aren't necessarily followed by a nul terminator.
26347 (handle_char_store): Likewise.
26349 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26351 PR tree-optimization/80769
26352 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
26353 for malloc and calloc. Document the new invariant that all related
26354 strinfos have delayed lengths or none do.
26355 (verify_related_strinfos): Move earlier in file.
26356 (set_endptr_and_length): New function, split out from...
26357 (get_string_length): ...here. Also set the lengths of related
26359 (zero_length_string): Assert that chainsi has known (rather than
26361 (adjust_related_strinfos): Likewise.
26363 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
26365 PR tree-optimization/81136
26366 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
26367 assert that two references with the same misalignment have the same
26368 compile-time misalignment if those compile-time misalignments
26371 2017-07-01 Andi Kleen <ak@linux.intel.com>
26373 * print-tree.c (print_node): Print all attributes.
26375 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26377 * cfg.c (scale_bbs_frequencies): New function.
26378 * cfg.h (scale_bbs_frequencies): Declare it.
26379 * cfgloopanal.c (single_likely_exit): Cleanup.
26380 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
26382 (scale_loop_profile): Likewise.
26383 (loop_version): Likewise.
26384 (create_empty_loop_on_edge): Update.
26385 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
26386 scale_loop_frequencies, scale_loop_profile, loopify,
26387 loop_version): Update prototypes.
26388 * modulo-sched.c (sms_schedule): Update.
26389 * predict.c (unlikely_executed_edge_p): Also check probability.
26390 (probably_never_executed_edge_p): Fix typo.
26391 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26392 * tree-parloops.c (gen_parallel_loop): Update.
26393 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
26394 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26395 * tree-ssa-loop-split.c (split_loop): Update.
26396 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26397 * tree-vect-loop-manip.c (vect_do_peeling): Update.
26398 (vect_loop_versioning): Update.
26399 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26401 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26403 * trans-mem.c (split_bb_make_tm_edge): Update profile.
26405 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26407 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
26408 to keep profile consistent.
26410 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26412 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
26413 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
26414 * profile-count.h (max_safe_multiplier): Make unsigned.
26415 (profile_count::guessed_zero): New.
26417 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26419 * bb-reorder.c (fix_up_crossing_landing_pad,
26420 fix_crossing_conditional_branches): Use make_single_succ_edge
26421 to keep profile consistent.
26423 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
26425 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
26428 2017-07-01 Jakub Jelinek <jakub@redhat.com>
26431 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
26432 the right scopes, make sure cond_jump isn't preserved between multiple
26433 iterations. Search for fallthru edge whenever there are 3+ edges and
26434 use find_fallthru_edge for it.
26436 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26438 Patch by Alexander Monakov <amonakov@ispras.ru>
26439 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
26440 probabilities consistently.
26442 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26444 * pa.c (pa_expand_compare_and_swap_loop): Update call of
26445 emit_cmp_and_jump_insns.
26447 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26450 * tree-inline.c (expand_call_inline): Combine profile statuses.
26452 2017-06-30 Andrew Pinski <apinski@cavium.com>
26454 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
26455 fold_stmt returned true.
26457 2017-06-30 Nathan Sidwell <nathan@acm.org>
26459 * ggc.h (empty_string): Delete.
26460 * cfgexpand.c (expand_asm_stmt): Use plain "".
26461 * optabs.c (expand_asm_memory_barrier): Likewise.
26462 * stringpool.c (empty_string): Delete.
26463 (digit_vector, digit_string): Delete.
26464 (ggc_alloc_string): Use plain "", don't optimize single digit
26465 strings. Use ggc_alloc_atomic.
26467 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
26469 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
26470 comparison set and one other set, use the cost of the non-comparison
26473 2017-06-30 Nathan Sidwell <nathan@acm.org>
26475 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
26478 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
26480 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
26481 loops. Remove now unneeded calls to gimple_switch_set_label() that
26482 just set removed labels to NULL_TREE.
26484 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
26486 * tree-ssanames.c (set_range_info_raw): Abstract from ...
26487 (set_range_info): ...here. Only call set_range_info_raw if domain
26489 (set_nonzero_bits): Call set_range_info_raw.
26490 * tree-ssanames.h (set_range_info_raw): New.
26492 2017-06-30 Jakub Jelinek <jakub@redhat.com>
26495 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
26496 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
26497 of nonimmediate_operand and <store_mask_constraint> instead of m
26498 for the input operand. For V8FI iterator, always split if input
26499 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
26500 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
26501 <store_mask_predicate> instead of register_operand and
26502 <store_mask_constraint> instead of v for the input operand. Make
26503 sure both operands aren't MEMs for if not <mask_applied>.
26505 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
26507 * lto-wrapper.c (copy_file) Close both file descriptors before
26510 2017-06-30 Martin Liska <mliska@suse.cz>
26513 * multiple_target.c (create_dispatcher_calls): Make ifunc
26514 also for function that don't have calls or are not referenced.
26516 2017-06-30 Richard Biener <rguenther@suse.de>
26518 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
26519 analyze the first scalar stmt. Move vector type computation
26520 for the BB case here from ...
26521 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
26522 live operation processing in the SLP case properly.
26524 2017-06-30 Richard Biener <rguenther@suse.de>
26526 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
26528 2017-06-30 Martin Liska <mliska@suse.cz>
26531 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
26532 before BUILT_IN_UNWIND_RESUME when ASAN is used.
26534 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
26536 * doc/invoke.texi (AArch64): Add missing options and remove redundant
26539 2017-06-30 Richard Biener <rguenther@suse.de>
26541 PR tree-optimization/81249
26542 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
26543 condition reduction result to original scalar type.
26545 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26547 * profile-count.h (enum profile_quality): Fix typos and whitespace
26550 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26552 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
26553 type for branch probabilities.
26555 2017-06-29 Julian Brown <julian@codesourcery.com>
26556 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26558 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
26559 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
26560 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
26561 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
26563 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26565 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
26566 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
26567 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
26568 CC usage from generic code to here.
26569 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
26570 CC usage into the target macros.
26572 2017-06-29 Maya Rashish <coypu@sdf.org>
26574 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
26577 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26579 * arm/arm-builtins.c: Include profile-count.h
26580 * except.c (sjlj_emit_function_enter): Use
26581 profile_probability::unlikely.
26583 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26585 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
26586 and tocrel_offset be pointer args rather than implicitly using
26588 (legitimate_constant_pool_address_p, rs6000_emit_move,
26589 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
26590 tocrel_offset and use in toc_relative_expr_p call.
26591 (print_operand, print_operand_address): Use static tocrel_base_oac
26592 and tocrel_offset_oac.
26593 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
26596 2017-06-29 Maya Rashish <coypu@sdf.org>
26598 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
26600 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
26602 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
26603 objects, take into account only the alignment of 'op0' and 'mode1' if
26606 2017-06-29 Steve Ellcey <sellcey@cavium.com>
26608 * ccmp.c (ccmp_tree_comparison_p): New function.
26609 (ccmp_candidate_p): Update to use above function.
26610 (get_compare_parts): New function.
26611 (expand_ccmp_next): Update to use new functions.
26612 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
26614 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
26615 take mode as argument.
26616 * ccmp.h (expand_ccmp_expr): Add mode as argument.
26617 * expr.c (expand_expr_real_1): Pass mode as argument.
26619 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
26621 * combine.c (combine_instructions): Print insns to dump_file, together
26624 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
26626 * asan.c (asan_emit_stack_protection): Update.
26627 (create_cond_insert_point): Update.
26628 * auto-profile.c (afdo_propagate_circuit): Update.
26629 * basic-block.h (struct edge_def): Turn probability to
26630 profile_probability.
26631 (EDGE_FREQUENCY): Update.
26632 * bb-reorder.c (find_traces_1_round): Update.
26633 (better_edge_p): Update.
26634 (sanitize_hot_paths): Update.
26635 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
26636 (make_single_succ_edge): Update.
26637 (check_bb_profile): Update.
26638 (dump_edge_info): Update.
26639 (update_bb_profile_for_threading): Update.
26640 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
26641 probabilitycount to 0.
26642 * cfgbuild.c (compute_outgoing_frequencies): Update.
26643 * cfgcleanup.c (try_forward_edges): Update.
26644 (outgoing_edges_match): Update.
26645 (try_crossjump_to_edge): Update.
26646 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
26647 (expand_gimple_tailcall): Update.
26648 (construct_init_block): Use make_single_succ_edge.
26649 (construct_exit_block): Use make_single_succ_edge.
26650 * cfghooks.c (verify_flow_info): Update.
26651 (redirect_edge_succ_nodup): Update.
26652 (split_edge): Update.
26653 (account_profile_record): Update.
26654 * cfgloopanal.c (single_likely_exit): Update.
26655 * cfgloopmanip.c (scale_loop_profile): Update.
26656 (set_zero_probability): Remove.
26657 (duplicate_loop_to_header_edge): Update.
26658 * cfgloopmanip.h (loop_version): Update prototype.
26659 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
26660 (force_nonfallthru_and_redirect): Update.
26661 (update_br_prob_note): Update.
26662 (rtl_verify_edges): Update.
26663 (purge_dead_edges): Update.
26664 (rtl_lv_add_condition_to_bb): Update.
26665 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
26666 * cgraphunit.c (init_lowered_empty_function): Update.
26667 (cgraph_node::expand_thunk): Update.
26668 * cilk-common.c: Include profile-count.h
26669 * dojump.c (inv): Remove.
26670 (jumpifnot): Update.
26671 (jumpifnot_1): Update.
26672 (do_jump_1): Update.
26674 (do_jump_by_parts_greater_rtx): Update.
26675 (do_compare_rtx_and_jump): Update.
26676 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
26677 do_jump_1. do_compare_rtx_and_jump): Update prototype.
26678 * dwarf2cfi.c: Include profile-count.h
26679 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
26680 (sjlj_emit_dispatch_table): Likewise.
26681 * explow.c: Include profile-count.h
26682 * expmed.c (emit_store_flag_force): Update.
26683 (do_cmp_and_jump): Update.
26684 * expr.c (compare_by_pieces_d::generate): Update.
26685 (compare_by_pieces_d::finish_mode): Update.
26686 (emit_block_move_via_loop): Update.
26687 (store_expr_with_bounds): Update.
26688 (store_constructor): Update.
26689 (expand_expr_real_2): Update.
26690 (expand_expr_real_1): Update.
26691 * expr.h (try_casesi, try_tablejump): Update prototypes.
26692 * gimple-pretty-print.c (dump_probability): Update.
26693 (dump_profile): New.
26694 (dump_gimple_label): Update.
26695 (dump_gimple_bb_header): Update.
26696 * graph.c (draw_cfg_node_succ_edges): Update.
26697 * hsa-gen.c (convert_switch_statements): Update.
26698 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
26699 (find_if_case_1): Update.
26700 (find_if_case_2): Update.
26701 * internal-fn.c (expand_arith_overflow_result_store): Update.
26702 (expand_addsub_overflow): Update.
26703 (expand_neg_overflow): Update.
26704 (expand_mul_overflow): Update.
26705 (expand_vector_ubsan_overflow): Update.
26706 * ipa-cp.c (good_cloning_opportunity_p): Update.
26707 * ipa-split.c (split_function): Use make_single_succ_edge.
26708 * ipa-utils.c (ipa_merge_profiles): Update.
26709 * loop-doloop.c (add_test): Update.
26710 (doloop_modify): Update.
26711 * loop-unroll.c (compare_and_jump_seq): Update.
26712 (unroll_loop_runtime_iterations): Update.
26713 * lra-constraints.c (lra_inheritance): Update.
26714 * lto-streamer-in.c (input_cfg): Update.
26715 * lto-streamer-out.c (output_cfg): Update.
26716 * mcf.c (adjust_cfg_counts): Update.
26717 * modulo-sched.c (sms_schedule): Update.
26718 * omp-expand.c (expand_omp_for_init_counts): Update.
26719 (extract_omp_for_update_vars): Update.
26720 (expand_omp_ordered_sink): Update.
26721 (expand_omp_for_ordered_loops): Update.
26722 (expand_omp_for_generic): Update.
26723 (expand_omp_for_static_nochunk): Update.
26724 (expand_omp_for_static_chunk): Update.
26725 (expand_cilk_for): Update.
26726 (expand_omp_simd): Update.
26727 (expand_omp_taskloop_for_outer): Update.
26728 (expand_omp_taskloop_for_inner): Update.
26729 * omp-simd-clone.c (simd_clone_adjust): Update.
26730 * optabs.c (expand_doubleword_shift): Update.
26731 (expand_abs): Update.
26732 (emit_cmp_and_jump_insn_1): Update.
26733 (expand_compare_and_swap_loop): Update.
26734 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
26735 * predict.c (predictable_edge_p): Update.
26736 (edge_probability_reliable_p): Update.
26737 (set_even_probabilities): Update.
26738 (combine_predictions_for_insn): Update.
26739 (combine_predictions_for_bb): Update.
26740 (propagate_freq): Update.
26741 (estimate_bb_frequencies): Update.
26742 (force_edge_cold): Update.
26743 * profile-count.c (profile_count::dump): Add missing space into dump.
26744 (profile_count::debug): Add newline.
26745 (profile_count::differs_from_p): Explicitly convert to unsigned.
26746 (profile_count::stream_in): Update.
26747 (profile_probability::dump): New member function.
26748 (profile_probability::debug): New member function.
26749 (profile_probability::differs_from_p): New member function.
26750 (profile_probability::differs_lot_from_p): New member function.
26751 (profile_probability::stream_in): New member function.
26752 (profile_probability::stream_out): New member function.
26753 * profile-count.h (profile_count_quality): Rename to ...
26754 (profile_quality): ... this one.
26755 (profile_probability): New.
26756 (profile_count): Update.
26757 * profile.c (compute_branch_probabilities): Update.
26758 * recog.c (peep2_attempt): Update.
26759 * sched-ebb.c (schedule_ebbs): Update.
26760 * sched-rgn.c (find_single_block_region): Update.
26761 (compute_dom_prob_ps): Update.
26762 (schedule_region): Update.
26763 * sel-sched-ir.c (compute_succs_info): Update.
26764 * stmt.c (struct case_node): Update.
26765 (do_jump_if_equal): Update.
26766 (get_outgoing_edge_probs): Update.
26767 (conditional_probability): Update.
26768 (emit_case_dispatch_table): Update.
26769 (expand_case): Update.
26770 (expand_sjlj_dispatch_table): Update.
26771 (emit_case_nodes): Update.
26772 * targhooks.c: Update.
26773 * tracer.c (better_p): Update.
26774 (find_best_successor): Update.
26775 * trans-mem.c (expand_transaction): Update.
26776 * tree-call-cdce.c: Update.
26777 * tree-cfg.c (gimple_split_edge): Upate.
26778 (move_sese_region_to_fn): Upate.
26779 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
26780 * tree-eh.c (lower_resx): Upate.
26781 (cleanup_empty_eh_move_lp): Upate.
26782 * tree-if-conv.c (version_loop_for_if_conversion): Update.
26783 * tree-inline.c (copy_edges_for_bb): Update.
26784 (copy_cfg_body): Update.
26785 * tree-parloops.c (gen_parallel_loop): Update.
26786 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
26787 (gimple_gen_time_profiler): Update.
26788 * tree-ssa-dce.c (remove_dead_stmt): Update.
26789 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
26790 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
26791 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
26792 (unloop_loops): Update.
26793 (try_peel_loop): Update.
26794 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
26795 * tree-ssa-loop-split.c (connect_loops): Update.
26796 (split_loop): Update.
26797 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
26798 (hoist_guard): Update.
26799 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
26800 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
26801 (value_replacement): Update.
26802 * tree-ssa-reassoc.c (branch_fixup): Update.
26803 * tree-ssa-tail-merge.c (replace_block_by): Update.
26804 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
26805 (create_edge_and_update_destination_phis): Update.
26806 (compute_path_counts): Update.
26807 (recompute_probabilities): Update.
26808 (update_joiner_offpath_counts): Update.
26809 (freqs_to_counts_path): Update.
26810 (duplicate_thread_path): Update.
26811 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
26812 (struct switch_conv_info): Update.
26813 (gen_inbound_check): Update.
26814 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
26815 (vect_do_peeling): Update.
26816 (vect_loop_versioning): Update.
26817 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
26818 (optimize_mask_stores): Update.
26819 * ubsan.c (ubsan_expand_null_ifn): Update.
26820 * value-prof.c (gimple_divmod_fixed_value): Update.
26821 (gimple_divmod_fixed_value_transform): Update.
26822 (gimple_mod_pow2): Update.
26823 (gimple_mod_pow2_value_transform): Update.
26824 (gimple_mod_subtract): Update.
26825 (gimple_mod_subtract_transform): Update.
26826 (gimple_ic): Update.
26827 (gimple_stringop_fixed_value): Update.
26828 (gimple_stringops_transform): Update.
26829 * value-prof.h: Update.
26831 2017-06-29 Carl Love <cel@us.ibm.com>
26833 * config/rs6000/rs6000-c.c: Add support for built-in functions
26834 vector signed int vec_signed (vector float);
26835 vector signed long long vec_signed (vector double);
26836 vector signed int vec_signed2 (vector double, vector double);
26837 vector signed int vec_signede (vector double);
26838 vector signed int vec_signedo (vector double);
26839 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
26840 instruction generator.
26841 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26842 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
26843 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
26845 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
26846 vunsignede_v2df): Add define_expands.
26847 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
26848 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
26849 VEC_UNSIGNEDO): Add definitions.
26850 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
26851 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
26852 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
26853 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
26854 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
26855 * config/rs6000/altivec.h (vec_signed, vec_signed2,
26856 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
26857 vec_unsignede, vec_unsignedo): Add builtin defines.
26858 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
26860 * doc/extend.texi: Update the built-in documentation file for the
26861 new built-in functions.
26863 2017-06-29 Richard Biener <rguenther@suse.de>
26865 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
26866 reduction chains to LOOP_VINFO_REDUCTIONS.
26867 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
26868 SLP reductions after processing reduction chains.
26870 2017-06-29 Nathan Sidwell <nathan@acm.org>
26872 * builtins.c (fold_builtin_FUNCTION): Use
26873 lang_hooks.decl_printable_name.
26875 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
26877 PR middle-end/81194
26878 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
26879 with only one label.
26880 * stmt.c (expand_case): Assert NCASES is greater than one.
26882 2017-06-29 Richard Biener <rguenther@suse.de>
26884 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
26886 (group_case_labels): Likewise.
26887 (find_taken_edge): Push sanity checking on val to workers...
26888 (find_taken_edge_cond_expr): ... here
26889 (find_taken_edge_switch_expr): ... and here, handle cases
26890 with just a default label.
26891 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
26892 (group_case_labels): Likewise.
26893 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
26894 group_case_labels does anything cleanup the CFG again.
26896 2017-06-29 Bin Cheng <bin.cheng@arm.com>
26898 PR tree-optimization/81196
26899 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
26900 exit condition comparing two IVs.
26902 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
26904 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
26905 profile to the dummy entry at the end of the list of architectures.
26906 * config/arm/arm-cpu-cdata.h: Regenerated.
26908 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26909 Michael Collison <michael.collison@arm.com>
26912 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
26914 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
26915 (*aarch64_reg_<mode>3_minus_mask): New pattern.
26916 (*aarch64_<optab>_reg_di3_mask2): New pattern.
26917 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
26918 of shift when the shift amount is masked with constant equal to
26919 the size of the mode.
26920 * config/aarch64/predicates.md (subreg_lowpart_operator): New
26923 2017-06-29 Martin Liska <mliska@suse.cz>
26925 * config/i386/i386.opt: Change range from [1,5] to [0,5].
26927 2017-06-29 Yury Gribov <tetra2005@gmail.com>
26930 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
26933 (edge_growth_cache_entry::edge_growth_cache_entry): New
26935 (reset_edge_growth_cache): Update to use constructor.
26937 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26939 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
26940 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26941 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
26943 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
26945 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
26946 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
26948 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
26950 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
26951 (*-linux-uclibc*): Add t-uclibc tmake_file.
26952 * config/t-musl: New.
26953 * config/t-uclibc: New.
26955 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
26957 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
26959 (gen_comm_data): Emit architectural setting of arch_prof.
26960 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
26962 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
26963 (armv8-m.base, armv8-m.main): Likewise.
26964 * arm-protos.h (arm_build_target): Add profile field.
26965 (arch_option): Likewise.
26966 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
26968 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
26969 arm_active_target.profile.
26971 2017-06-28 Richard Biener <rguenther@suse.de>
26973 PR middle-end/81227
26974 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
26975 TYPE_OVERFLOW_WRAPS.
26976 * match.pd (negate_expr_p): Likewise.
26977 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
26978 fold_build2, not fold_binary.
26980 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26982 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
26983 Convert memory address to Pmode.
26984 (aarch64_print_operand): Assert MEM operands are always Pmode.
26986 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
26989 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
26990 Remove redundant if.
26991 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
26992 * config/arm/aarch-common-protos.h
26993 (aarch_forward_to_shift_is_not_shifted_re): Remove.
26994 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
26996 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
26999 * multiple_target.c (create_dispatcher_calls): Set the default
27000 clone to be static, not public.
27002 2017-06-28 Richard Biener <rguenther@suse.de>
27004 * tree-vect-loop.c (vectorizable_reduction): Move special
27005 cond reduction IV var creation ...
27006 (vect_create_epilog_for_reduction): ... here. Remove induction_index
27007 parameter. Use STMT_VINFO_VECTYPE.
27008 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
27011 2017-06-28 Martin Liska <mliska@suse.cz>
27014 * ipa-visibility.c (non_local_p): Handle visibility.
27016 2017-06-28 Martin Liska <mliska@suse.cz>
27019 * common.opt: Add IntegerRange to various options.
27020 * opt-functions.awk (integer_range_info): New function.
27021 * optc-gen.awk: Add integer_range_info to cl_options struct.
27022 * opts-common.c (decode_cmdline_option): Handle
27023 CL_ERR_INT_RANGE_ARG.
27024 (cmdline_handle_error): Likewise.
27025 * opts.c (print_filtered_help): Show valid interval in
27026 when --help is provided.
27027 * opts.h (struct cl_option): Add range_min and range_max fields.
27028 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
27030 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
27032 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
27033 (x * C EQ/NE y * C): New transformation.
27035 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
27037 * genmultilib (combination_space): Accept '+' in option names.
27039 2017-06-28 Martin Liska <mliska@suse.cz>
27042 * asan.c (instrument_derefs): Bail out inner references
27043 that are hard register variables.
27045 2017-06-28 Jakub Jelinek <jakub@redhat.com>
27048 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
27049 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
27051 2017-06-28 Richard Biener <rguenther@suse.de>
27053 * tree-vectorizer.h (vect_get_vec_defs): Remove.
27054 (vect_get_slp_defs): Adjust.
27055 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
27057 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
27059 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
27060 get_initial_defs_for_reduction instead of vect_get_vec_defs.
27061 (vectorizable_reduction): Adjust.
27062 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
27064 (vect_get_slp_defs): Likewise.
27065 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
27066 (vectorizable_bswap): Adjust.
27067 (vectorizable_call): Likewise.
27068 (vectorizable_conversion): Likewise.
27069 (vectorizable_assignment): Likewise.
27070 (vectorizable_shift): Likewise.
27071 (vectorizable_operation): Likewise.
27072 (vectorizable_store): Likewise.
27073 (vectorizable_condition): Likewise.
27074 (vectorizable_comparison): Likewise.
27076 2017-06-28 Michael Collison <michael.collison@arm.com>
27079 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
27081 (arm_gen_movmemqi): Removed unused variable 'i'.
27082 Convert 'for' loop into 'while' loop.
27083 (arm_expand_prologue): Remove last unnecessary set of insn.
27084 (thumb_pop): Remove unused variable 'pushed_words'.
27085 (thumb_exit): Remove last unnecessary set of regs_to_pop.
27087 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27089 * config/s390/predicates.md: Use s390_rel_address_ok_p.
27090 * config/s390/s390-protos.h: Add prototype of
27091 s390_rel_address_ok_p.
27092 * config/s390/s390.c (s390_got_symbol): New function.
27093 (s390_rel_address_ok_p): New function.
27094 (legitimize_pic_address): Use s390_rel_address_ok_p.
27095 (s390_load_got): Use s390_got_symbol.
27096 (s390_option_override): Issue error if
27097 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
27098 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
27100 * config/s390/s390.opt: New option mpic-data-is-text-relative.
27102 2017-06-27 Andrew Pinski <apinski@cavium.com>
27104 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
27105 (X * copysign (1.0, X)): New pattern.
27106 (X * copysign (1.0, -X)): New pattern.
27107 (copysign (-1.0, CST)): New pattern.
27109 2017-06-27 Joseph Myers <joseph@codesourcery.com>
27111 * genmultilib (combination_space): Remove variable.
27112 Validate reuse rules against regular expression for any sequence
27113 of multilib options in any order.
27115 2017-06-27 Michael Collison <michael.collison@arm.com>
27117 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
27118 call aarch64_split_simd_combine.
27119 * (aarch64_combine_internal<mode>): Delete pattern.
27120 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
27121 Allow register and subreg operands.
27123 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27125 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
27126 specific need, just fallback on defaults.
27127 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
27129 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27130 Olivier Hainque <hainque@adacore.com>
27132 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
27134 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
27135 targets. Pick a default if no particular attempt applied.
27136 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
27139 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27141 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
27142 (x86_64-wrs-vxworks7): Likewise.
27144 2017-06-27 Marek Polacek <polacek@redhat.com>
27147 * ubsan.c (instrument_null): Check get_base_address's result for null.
27149 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27151 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
27153 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
27155 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
27156 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
27157 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
27158 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
27159 New function types.
27160 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
27161 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
27162 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
27163 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
27164 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
27165 BUILT_IN_FEUPDATEENV): New builtins.
27166 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
27167 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
27168 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
27169 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
27171 (builtin_structptr_types): Adjust size.
27172 * tree.c (builtin_structptr_types): Add four entries.
27174 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27175 Olivier Hainque <hainque@adacore.com>
27177 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
27178 (TLS_SYM): New local macro, forcing reference to __tls__ on
27179 link command lines for VxWorks 7 RTPs, triggering initialization
27181 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
27182 OS features TLS support, true for RTPs on VxWorks 7.
27183 * config/vxworks.c (vxworks_override_options): Setup emutls
27186 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27188 * predict.c (test_prediction_value_range): Use -1U instead of -1
27189 to avoid narrowing conversion warning.
27190 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
27191 to avoid narrowing conversion warning.
27192 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
27194 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
27196 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
27198 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
27199 64bit configurations.
27200 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
27201 (SIZE_TYPE): Likewise.
27202 * config/vxworks.c (vxworks_emutls_var_fields): Use
27203 long_unsigned_type_node instead of unsigned_type_node as the offset
27204 field type, which is "pointer" mode in emutls.c.
27206 2017-06-27 Jakub Jelinek <jakub@redhat.com>
27209 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
27211 PR middle-end/81207
27212 * gimple-fold.c (replace_call_with_call_and_fold): Handle
27213 gimple_vuse copying separately from gimple_vdef copying.
27215 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
27217 * value-prof.c (free_hist): Remove call to memset and the enclosing if
27220 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27221 Olivier Hainque <hainque@adacore.com>
27223 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
27224 for all vxworks7 targets.
27225 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
27226 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
27227 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
27228 variations for VX6/VX7 and 32/64bits later on in ...
27229 (VXWORKS_LIB_SPEC): Leverage new macros.
27230 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
27231 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
27233 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
27235 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
27236 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
27238 2017-06-26 Carl Love <cel@us.ibm.com>
27240 * config/rs6000/rs6000-c.c: Add support for built-in functions
27241 vector bool char vec_reve (vector bool char);
27242 vector signed char vec_reve (vector signed char);
27243 vector unsigned char vec_reve (vector unsigned char);
27244 vector bool int vec_reve (vector bool int);
27245 vector signed int vec_reve (vector signed int);
27246 vector unsigned int vec_reve (vector unsigned int);
27247 vector bool long long vec_reve (vector bool long long);
27248 vector signed long long vec_reve (vector signed long long);
27249 vector unsigned long long vec_reve (vector unsigned long long);
27250 vector bool short vec_reve (vector bool short);
27251 vector signed short vec_reve (vector signed short);
27252 vector double vec_reve (vector double);
27253 vector float vec_reve (vector float);
27254 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
27255 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
27256 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
27257 (altivec_vreve): New pattern.
27258 * config/rs6000/altivec.h (vec_reve): New define.
27259 * doc/extend.texi (vec_rev): Update the built-in documentation file
27260 for the new built-in functions.
27262 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27264 PR tree-optimization/71815
27265 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
27267 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
27269 (slsr_process_phi): Likewise.
27270 (replace_uncond_cands_and_profitable_phis): Don't replace a
27271 multiply candidate with a stride of 1 (copy or cast).
27272 (phi_incr_cost): Call uses_consumed_by_stmt rather than
27274 (lowest_cost_path): Likewise.
27275 (total_savings): Likewise.
27277 2017-06-26 Richard Biener <rguenther@suse.de>
27280 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
27281 Use def_builtin_pure for all gather builtins.
27283 2017-06-26 Richard Biener <rguenther@suse.de>
27285 PR tree-optimization/81203
27286 * tree-tailcall.c (find_tail_calls): Do not move stmts into
27287 non-dominating BBs.
27289 2017-06-26 Marek Polacek <polacek@redhat.com>
27292 * doc/invoke.texi: Document -Wmultistatement-macros.
27294 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
27296 * doc/sourcebuild.texi (ARM-specific attributes): Document new
27297 arm_neon_ok_no_float_abi effective target.
27299 2017-06-26 Richard Biener <rguenther@suse.de>
27301 PR tree-optimization/80928
27302 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
27303 (copy_bbs): Set BB_DUPLICATED flag early.
27304 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
27306 (execute_on_shrinking_pred): Likewise.
27307 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
27308 BB_DUPLICATED blocks.
27309 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
27310 iterate over all PHIs considering removal of *gsi.
27312 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
27314 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
27317 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27319 * config/rs6000/rs6000-string.c: (expand_block_clear,
27320 do_load_for_compare, select_block_compare_mode,
27321 compute_current_alignment, expand_block_compare,
27322 expand_strncmp_align_check, expand_strn_compare,
27323 expand_block_move, rs6000_output_load_multiple)
27324 Move functions related to string/block move/compare
27325 to a separate file.
27326 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
27327 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
27328 for this function which is now used in two files.
27329 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
27330 * config.gcc: Add rs6000-string.o to extra_objs for
27331 targets powerpc*-*-* and rs6000*-*-*.
27333 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
27336 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
27337 32-bit, since indexed is not valid for DImode.
27338 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
27339 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
27340 (define_peephole2 for Altivec d-form load): Add 32-bit support.
27341 (define_peephole2 for Altivec d-form store): Likewise.
27344 * multiple_target.c (create_dispatcher_calls): Only create the
27345 dispatcher call if the function is the default clone of a
27346 versioned function.
27348 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
27350 PR middle-end/80902
27351 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
27352 a call, force the call to not be a tail call.
27354 2017-06-23 Jeff Law <law@redhat.com>
27356 * doc/contrib.texi: Add entry for Steven Pemberton's work on
27359 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
27361 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
27362 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
27363 handling for early expansion of vector shifts (sl,sr,sra,rl).
27364 (builtin_function_type): Add vector shift right instructions
27365 to the unsigned argument list.
27367 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
27369 rtl-optimizatoin/79286
27370 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
27371 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
27372 trap. PIC register plus a const unspec without offset can never trap.
27374 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
27376 * tree.h (builtin_structptr_type): New type.
27377 (builtin_structptr_types): Declare new array.
27378 * tree.c (builtin_structptr_types): New array.
27379 (free_lang_data, build_common_tree_nodes): Use it.
27381 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
27384 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
27387 2017-06-22 Matt Turner <mattst88@gmail.com>
27389 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
27390 Lake models to skylake case. Assume skylake for unknown
27391 models with clflushopt.
27393 2017-06-22 Jeff Law <law@redhat.com>
27395 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
27396 frame sizes that do not satisfy aarch64_uimm12_shift.
27398 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
27400 * profile-count.h (apply_probability,
27401 apply_scale, probability_in): Fix checks for zero.
27403 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27405 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
27406 * doc/cppdiropts.texi (-I @var{dir}): Document it.
27408 2016-06-22 Richard Biener <rguenther@suse.de>
27410 * tree-vect-loop.c (vect_model_reduction_cost): Handle
27411 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
27412 REDUC_MAX_EXPR support.
27413 (vectorizable_reduction): Likewise.
27414 (vect_create_epilog_for_reduction): Likewise.
27416 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
27418 * match.pd (A / (1 << B) -> A >> B): New.
27419 * generic-match-head.c: Include optabs-tree.h.
27420 * gimple-match-head.c: Likewise.
27421 * optabs-tree.h (target_supports_op_p): New.
27422 * optabs-tree.c (target_supports_op_p): New.
27424 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27426 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
27427 $gcc_cv_ld --help output.
27428 (gcc_cv_ld_demangle): Likewise.
27429 (gcc_cv_ld_eh_frame_hdr): Likewise.
27430 (gcc_cv_ld_pie): Likewise.
27431 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
27432 (gcc_cv_ld_buildid): Likewise.
27433 (gcc_cv_ld_sysroot): Likewise.
27434 (ld_bndplt_support): Likewise.
27435 (ld_pushpopstate_support): Likewise.
27436 * configure: Regenerate.
27437 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
27439 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27442 * config/i386/sse.md (round<mode>2): Renumber match_dup and
27443 operands indexes to avoid gap between operands and match_dups.
27445 2017-06-21 Andrew Pinski <apinski@cavium.com>
27447 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
27448 Increment Arith_shift and Arith_shift_reg by 1.
27449 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
27451 * config/aarch64/aarch64.c (thunderx_tunings): Enable
27452 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
27453 (aarch64_strip_extend): Add new argument and test for it.
27454 (aarch64_cheap_mult_shift_p): New function.
27455 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
27456 add a cost if it is true.
27457 Update calls to aarch64_strip_extend.
27458 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
27460 2017-06-21 Andrew Pinski <apinski@cavium.com>
27462 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
27464 (thunderxt88): Likewise.
27465 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
27466 (thunderx_prefetch_tune): New variable.
27467 (thunderx2t99_prefetch_tune): Update for the correct values.
27468 (thunderxt88_tunings): New variable.
27469 (thunderx_tunings): Use thunderx_prefetch_tune instead of
27470 generic_prefetch_tune.
27471 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
27473 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27475 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
27476 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
27477 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
27478 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
27479 (aarch64_atomic_cas<mode>, GPI): Likewise.
27481 2017-06-21 Martin Liska <mliska@suse.cz>
27483 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
27484 statements on cold and hot labels.
27485 * predict.c (tree_estimate_probability_bb): Remove the
27486 prediction from this place.
27488 2017-06-21 Martin Liska <mliska@suse.cz>
27490 PR tree-optimization/79489
27491 * gimplify.c (maybe_add_early_return_predict_stmt): New
27493 (gimplify_return_expr): Call the function.
27494 * predict.c (tree_estimate_probability_bb): Remove handling
27496 * predict.def: Update comment about early return predictor.
27497 * gimple-predict.h (is_gimple_predict): New function.
27498 * predict.def: Change default value of early return to 66.
27499 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
27501 * passes.def: Put pass_strip_predict_hints to the beginning of
27504 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
27506 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
27507 FUNCTION_DECL declarations.
27508 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
27510 (dwaf2out_decl): Likewise.
27511 * godump.c (go_early_global_decl): Skip call to the real debug hook
27512 for FUNCTION_DECL declarations.
27513 * passes.c (rest_of_decl_compilation): Skip call to the
27514 early_global_decl debug hook for FUNCTION_DECL declarations, unless
27515 -fdump-go-spec is passed.
27517 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27519 * config/i386/i386.c (struct builtin_isa): New field pure_p.
27520 Reorder for compactness.
27521 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
27522 (def_builtin_pure, def_builtin_pure2): New functions.
27523 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
27525 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
27527 * match.pd (nop_convert): New predicate.
27528 ((A +- CST1) +- CST2): Allow some NOP conversions.
27530 2017-06-21 Jakub Jelinek <jakub@redhat.com>
27533 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
27534 with ctors/dtors if GOVD_SHARED is set.
27536 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27538 * config/aarch64/aarch64.md (movti_aarch64):
27539 Emit mov rather than orr.
27540 (movtf_aarch64): Likewise.
27541 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
27542 Emit mov rather than orr.
27544 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27546 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
27547 Swap alternatives, make integer dup more expensive.
27549 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
27551 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
27552 Return true for non-tls symbols.
27554 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
27556 * config/aarch64/aarch64-cores.def (cortex-a55): New.
27557 (cortex-a75): Likewise.
27558 (cortex-a75.cortex-a55): Likewise.
27559 * config/aarch64/aarch64-tune.md: Regenerate.
27560 * doc/invoke.texi (-mtune): Document new values for -mtune.
27562 2017-06-21 Tom de Vries <tom@codesourcery.com>
27564 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
27565 stack_size feature.
27566 (Effective-Target Keywords, Other attributes): Suggest using
27567 dg-add-options stack_size feature to get stack limit in stack_size
27568 effective target documentation.
27570 2017-06-21 Julian Brown <julian@codesourcery.com>
27571 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27573 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
27574 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
27575 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
27577 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
27578 attribute type list for neon_multiply.
27579 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
27580 attribute type list for neon_multiply.
27581 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
27582 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
27583 attribute type list for neon_multiply.
27584 * config/arm/types.md (crypto_pmull): Add.
27585 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
27586 attribute type list.
27588 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
27590 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
27593 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27595 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
27596 to make sure not to dereference a NULL cost_classes_ptr pointer.
27598 2017-06-20 Carl Love <cel@us.ibm.com>
27600 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27601 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
27602 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
27603 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
27604 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
27605 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
27606 VMULOSW): New enum "unspec" values.
27607 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
27608 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
27609 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
27610 altivec_vmulosw): New patterns.
27611 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
27612 VMULOSW): Add definitions.
27614 2017-06-20 Julia Koval <julia.koval@intel.com>
27616 * config/i386/i386.c: Fix rounding expand for new pattern.
27617 * config/i386/subst.md: Fix pattern (parallel -> unspec).
27619 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27621 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
27622 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
27624 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27626 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
27629 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
27631 * config/aarch64/aarch64-cores.def: Rearrange to sort by
27632 architecture, then by implementer ID.
27633 * config/aarch64/aarch64-tune.md: Regenerate.
27635 2017-06-20 Richard Biener <rguenther@suse.de>
27637 PR middle-end/81097
27638 * fold-const.c (split_tree): Fold to type before negating.
27640 2017-06-20 David Malcolm <dmalcolm@redhat.com>
27642 * diagnostic-show-locus.c
27643 (selftest::test_fixit_deletion_affecting_newline): New function.
27644 (selftest::diagnostic_show_locus_c_tests): Call it.
27646 2017-06-20 Andreas Schwab <schwab@suse.de>
27649 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
27652 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
27654 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
27655 __ARM_FEATURE_COPROC according to support.
27657 2017-06-20 Jakub Jelinek <jakub@redhat.com>
27659 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
27660 Rewritten to avoid overflow for > 32-bit pointers.
27663 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
27664 by removing enum keyword.
27665 (ubsan_type_descriptor): Likewise. Formatting fix.
27668 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
27669 splitter): Require TARGET_SSE2 in the condition.
27671 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
27674 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
27675 for doing vector set of SFmode on ISA 3.0.
27676 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
27677 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
27679 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
27680 SFmode value into a V4SF variable that was extracted from another
27681 V4SF variable without converting the element to double precision
27682 and back to single precision vector format.
27683 (vsx_insert_extract_v4sf_p9_2): Likewise.
27685 2017-06-19 Jakub Jelinek <jakub@redhat.com>
27687 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
27688 in UWHI to avoid undefined overflow.
27691 * ubsan.h (enum ubsan_encode_value_phase): New.
27692 (ubsan_encode_value): Change second argument to
27693 enum ubsan_encode_value_phase with default value of
27694 UBSAN_ENCODE_VALUE_GENERIC.
27695 * ubsan.c (ubsan_encode_value): Change second argument to
27696 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
27697 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
27698 create_tmp_var_raw instead of create_tmp_var and use a
27700 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
27701 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
27702 ubsan_encode_value callers.
27705 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
27706 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
27707 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
27709 2017-06-19 Richard Biener <rguenther@suse.de>
27711 PR middle-end/81118
27712 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
27713 estimates if we changed anything.
27715 2017-06-19 Richard Biener <rguenther@suse.de>
27717 PR tree-optimization/80887
27718 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
27719 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
27720 simplified lookups, then reset mprts_hook.
27721 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
27723 (try_to_simplify): Likewise.
27725 2017-06-19 Martin Liska <mliska@suse.cz>
27728 * gimplify.c (gimplify_switch_expr):
27729 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
27731 2017-06-19 Martin Liska <mliska@suse.cz>
27733 * doc/install.texi: Document that PGO runs in 4 stages.
27735 2017-06-19 Martin Liska <mliska@suse.cz>
27738 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
27739 to dispatcher function name.
27740 * multiple_target.c (replace_function_decl): New function.
27741 (create_dispatcher_calls): Redirect both edges and references.
27743 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
27745 * profile-count.c (profile_count::dump): Dump quality.
27746 (profile_count::differs_from_p): Update for unsigned val.
27747 * profile-count.h (profile_count_quality): New enum.
27748 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
27750 2017-06-19 Richard Biener <rguenther@suse.de>
27752 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
27753 struct function as arg.
27754 (estimate_numbers_of_iterations): Export overload with loop arg.
27755 (free_numbers_of_iterations_estimates_loop): Use an overload of
27756 free_numbers_of_iterations_estimates instead.
27757 * tree-cfg.c (remove_bb): Adjust.
27758 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
27759 * tree-parloops.c (gen_parallel_loop): Likewise.
27760 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
27762 (tree_unroll_loops_completely): Likewise.
27763 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
27764 Use an overload instead and export.
27765 (estimated_loop_iterations): Adjust.
27766 (max_loop_iterations): Likewise.
27767 (likely_max_loop_iterations): Likewise.
27768 (estimate_numbers_of_iterations): Take struct function as arg
27770 (loop_exits_before_overflow): Adjust.
27771 (free_numbers_of_iterations_estimates_loop): Use an overload.
27772 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
27773 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
27775 2017-06-19 Richard Biener <rguenther@suse.de>
27778 * ipa-prop.c (find_constructor_constant_at_offset): Handle
27779 RANGE_EXPR conservatively.
27781 2017-06-16 Carl Love <cel@us.ibm.com>
27783 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
27784 definitions for vec_float, vec_float2, vec_floato,
27785 vec_floate built-ins.
27786 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
27787 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
27789 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
27790 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
27791 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
27792 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
27793 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
27794 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
27795 vec_floato): Add builtin defines.
27796 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
27797 Update the built-in documentation file for the new built-in
27800 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27802 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
27803 (mthumb): Mark as the negative of -marm.
27805 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27807 * doc/invoke.texi (ARM Options, -mcpu): Document supported
27809 (ARM Options, -mtune): Document that this accepts the same
27810 extension options as -mcpu.
27811 (ARM Options, -mfpu): Document addition of -mfpu=auto.
27813 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27815 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
27816 permitted extensions.
27818 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27820 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
27821 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
27822 (armv8-m.main): Add option +nodsp.
27823 * config/arm/arm-cpu-cdata.h: Regenerated.
27825 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27827 * config/arm/t-fuchsia: New file.
27828 * config.gcc (arm*-*-fuchsia*): Use it.
27830 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27832 * config/arm/t-symbian: Rewrite for new option infrastructure.
27834 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27836 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
27837 (MULTILIB_REQUIRED): Likewise.
27839 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27841 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
27842 (MULTILIB_RESUE): Likewise.
27843 (MULTILIB_MATCHES): Likewise.
27844 (MULTLIB_REQUIRED): Likewise.
27846 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27848 * config/arm/t-rtems: Rewrite for new option framework.
27850 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27852 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
27853 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
27854 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
27855 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
27856 * config/arm/t-multilib: ... here.
27857 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
27858 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
27859 armv7-a and armv8*-a when A-profile libraries have not been built.
27860 * config/arm/t-rmprofile: Rewrite.
27862 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27864 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
27865 with a backslash. Remove the backslash after substituting unescaped
27867 * doc/fragments.texi (MULTILIB_REUSE): Document it.
27869 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27871 * config.gcc: (arm*-*-*): When building a-profile libraries, force
27872 the driver to pass through the default setting of -mfloat-abi.
27873 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
27875 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
27876 (all_feat_combs): New rule.
27877 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
27879 * config/arm/t-aprofile: Rewrite.
27881 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27883 * config/arm/arm.h (FPUTYPE_AUTO): Define.
27884 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
27885 fpu is not specified by the user/command-line.
27886 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
27887 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
27888 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
27889 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
27890 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
27891 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
27893 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27895 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
27896 * config/arm/t-arm-elf: Rewritten.
27898 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27900 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
27901 have some floating-point instructions.
27902 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
27903 (TARGET_MAYBE_HARD_FLOAT): New macro.
27904 * config/arm/arm-builtins.c (arm_init_builtins): Use
27905 TARGET_MAYBE_HARD_FLOAT.
27906 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
27908 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27910 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
27911 (configargs.h): Include it.
27912 (arm_print_hint_for_fpu_option): New function.
27913 (arm_parse_fpu_option): New function.
27914 (candidate_extension): New class.
27915 (arm_canon_for_multilib): New function.
27916 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
27917 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
27918 (ARCH_CANONICAL_SPECS): New macro.
27919 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
27921 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27923 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
27924 are set after handling multilib fragments. Set target_cpu_default2
27927 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27929 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
27931 (arm*-*-*): Set target_cpu_default2 to a quoted string.
27932 * config/arm/parsecpu.awk (check_cpu): Validate any extension
27934 (check_arch): Likewise.
27935 * config/arm/arm.c (arm_configure_build_target): Handle
27936 TARGET_CPU_DEFAULT being a string constant. Scan any feature
27937 options in the default.
27939 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27941 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
27942 when an option is an alias of another.
27943 * config/arm/parsecpu.awk (optalias): New parser token.
27944 (gen_comm_data): Mark non-alias options as such. Emit entries
27945 for extension aliases.
27946 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
27947 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
27948 (armv6kz, armv6zk, armv6t2): Likewise.
27949 (armv7): Make vfpv3-d16 an alias.
27950 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
27952 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
27953 Sort in canonical order.
27954 (armv8-a): Sort in canonical order.
27955 (armv8.1-a, armv8.2-a): Likewise.
27956 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
27958 (cortex-a9): Sort in canonical order.
27959 * config/arm/arm.c (selftests.h): Include it.
27960 (arm_test_cpu_arch_data): New function.
27961 (arm_run_self_tests): New function.
27962 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
27963 (targetm): Move declaration to the end of the file.
27964 * arm-cpu-cdata.h: Regenerated.
27966 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27968 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
27969 call to target_mode_check describing the type of option passed.
27970 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
27971 (arm_target_thumb_only): Use arm_parse_arch_option_name or
27972 arm_parse_cpu_option_name to match parameters against list of
27974 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
27975 arm_arch_core_flags data structure.
27976 * config/arm/arm-cpu_cdata.h: Regenerated.
27978 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
27980 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
27982 (arm_print_hint_for_cpu_option): Likewise.
27983 (arm_print_hint_for_arch_option): Likewise.
27984 (arm_parse_cpu_option_name): Likewise.
27985 (arm_parse_arch_option_name): Likewise.
27986 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
27987 of entries in the all_fpus list.
27988 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
27989 (arm_parse_cpu_option_name): Declare.
27990 (arm_parse_arch_option_name): Declare.
27991 (arm_parse_option_features): Declare.
27992 (arm_intialize_isa): Declare.
27993 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
27995 (gen_comm_data): ... here. Make definitions non-static.
27996 * config/arm/arm-cpu-data.h: Regenerated.
27997 * config/arm/arm-cpu-cdata.h: Regenerated.
27999 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28001 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
28002 (cpu_arch_extension): New structure.
28003 (cpu_arch_option, arch_option, cpu_option): New structures.
28004 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
28005 architecture types.
28006 (gen_data): Generate new format data tables.
28007 * config/arm/arm.c (cpu_tune): New structure.
28008 (cpu_option, processors): Delete.
28009 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
28010 (arm_print_hint_for_cpu_option): ... this and ...
28011 (arm_print_hint_for_arch_option): ... this.
28012 (arm_parse_arch_cpu_name): Delete. Replace with ...
28013 (arm_parse_cpu_option_name): ... this and ...
28014 (arm_parse_arch_option_name): ... this.
28015 (arm_unrecognized_feature): Change type of target parameter to
28017 (arm_parse_arch_cpu_features): Delete. Replace with ...
28018 (arm_parse_option_features): ... this.
28019 (arm_configure_build_target): Rework to use new configuration data
28021 (arm_print_tune_info): Rework for new configuration data tables.
28022 * config/arm/arm-cpu-data.h: Regenerated.
28023 * config/arm/arm-cpu.h: Regenerated.
28025 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28027 * Makefile.in (OBJS): Move sbitmap.o from here ...
28028 (OBJS-libcommon): ... to here.
28030 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28032 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
28033 (ISA_ALL_CRYPTO): New macro.
28034 (ISA_ALL_SIMD): New macro
28035 (ISA_ALL_FP): New macro.
28036 * config/arm/arm.c (fpu_bitlist): Update initializer.
28037 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
28039 (arm9e): Add fpu. Add option for nofp
28040 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
28041 (arm926ej-s, arm1026ej-s): Likewise.
28042 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
28043 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
28044 neon-fp16, neon-vfpv4, nofp and nosimd.
28045 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
28046 (cortex-a8): Add fpu. Add option for nofp.
28047 (cortex-a9): Add fpu. Add options for nosimd and nofp.
28048 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
28049 (cortex-r4f): Add fpu.
28050 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
28051 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
28053 (cortex-r8): Likewise.
28054 (cortex-m4): Add fpu. Add option for nofp.
28055 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
28056 (cortex-a17.cortex-a7): Likewise.
28057 (cortex-a32): Add fpu. Add options for crypto and nofp.
28058 (cortex-a35, cortex-a53): Likewise.
28059 (cortex-a57): Add fpu. Add option for crypto.
28060 (cortex-a72, cortex-a73): Likewise.
28061 (exynos-m1): Likewise.
28062 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
28063 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
28064 (cortex-m33): Add fpu. Add option for nofp.
28065 * config/arm/arm-cpu-cdata.h: Regenerated
28066 * config/arm/arm-cpu-data.h: Regenerated.
28068 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28070 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
28071 (armv5te, armv5tej): Likewise.
28072 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
28073 (armv7): Add options fp and vfpv3-d16.
28074 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
28075 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
28077 (armv7ve): Likewise.
28078 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
28079 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
28080 (armv8-a): Add nocrypto option.
28081 (armv8.1-a, armv8.2-a): Likewise.
28082 (armv8-m.main): add options fp, fp.dp and nofp.
28084 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28086 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
28088 (armv8-a+crc): Delete.
28089 (armv8.1-a): Add options simd, crypto and nofp.
28090 (armv8.2-a): Add options fp16, simd, crypto and nofp.
28091 (armv8.2-a+fp16): Delete.
28092 (armv8-m.main): Add option dsp.
28093 (armv8-m.main+dsp): Delete.
28094 (cortex-a8): Add fpu. Add nofp option.
28095 (cortex-a9): Add fpu. Add nofp and nosimd options.
28096 * config/arm/parsecpu.awk (gen_data): Generate option tables and
28097 link to main cpu and architecture data structures.
28098 (gen_comm_data): Only put isa attributes from the main architecture
28100 (option): New statement for architecture and CPU entries.
28101 * arm.c (struct cpu_option): New structure.
28102 (struct processors): Add entry for options.
28103 (arm_unrecognized_feature): New function.
28104 (arm_parse_arch_cpu_name): Ignore any characters after the first
28106 (arm_parse_arch_cpu_feature): New function.
28107 (arm_configure_build_target): Separate out any CPU and architecture
28108 features and parse separately. Don't error out if -mfpu=auto is
28109 used with only an architecture string.
28110 (arm_print_asm_arch_directives): New function.
28111 (arm_file_start): Call it.
28112 * config/arm/arm-cpu-cdata.h: Regenerated.
28113 * config/arm/arm-cpu-data.h: Likewise.
28114 * config/arm/arm-tables.opt: Likewise.
28116 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28118 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
28119 assembler when it is not -mfpu=auto.
28121 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28123 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
28124 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
28125 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
28126 (ASM_CPU_SPEC): Rewrite.
28127 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
28128 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
28129 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
28130 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
28131 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
28132 copied string is NUL-terminated. Also strip any characters prefixed
28134 (arm_rewrite_selected_arch): New function.
28135 (arm_rewrite_march): New function.
28137 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
28139 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
28140 (x_arm_cpu_string, x_arm_tune_string): Likewise.
28141 (march, mcpu, mtune): Convert to string-based options.
28142 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
28143 (arm_parse_arch_cpu_name): New function.
28144 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
28145 identify selected architecture or CPU.
28146 (arm_option_save): New function.
28147 (TARGET_OPTION_SAVE): Redefine.
28148 (arm_option_restore): Restore string options.
28149 (arm_option_print): Print string options.
28151 2017-06-16 Martin Sebor <msebor@redhat.com>
28153 PR tree-optimization/80933
28154 PR tree-optimization/80934
28155 * builtins.c (fold_builtin_3): Do not handle bcmp here.
28156 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
28157 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
28158 (gimple_fold_builtin): Call them.
28160 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28162 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
28163 as unlikely; update profile.
28165 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28167 * predict.c (force_edge_cold): Handle declaring edges impossible
28170 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28172 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
28174 (try_unroll_loop_completely): Fix reporting.
28176 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28178 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
28180 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
28183 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
28184 if given a non-constant argument for an intrinsic which requires a
28187 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28189 * profile.c (compare_freqs): New function.
28190 (branch_prob): Sort edge list.
28191 (find_spanning_tree): Assume that the list is priority sorted.
28193 2017-06-16 Richard Biener <rguenther@suse.de>
28195 PR tree-optimization/81090
28196 * passes.def (pass_record_bounds): Remove.
28197 * tree-pass.h (make_pass_record_bounds): Likewise.
28198 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
28199 make_pass_record_bounds): Likewise.
28200 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
28201 not free niter estimates at the beginning but at the end.
28202 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
28204 2017-06-16 Richard Biener <rguenther@suse.de>
28206 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
28207 initializer to workaround ICE in host GCC 4.8.
28209 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
28211 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
28213 (clone_inlined_nodes): Update.
28215 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28217 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
28218 prefetch settings, and enable prefetching by default at -O3.
28220 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28222 * config/aarch64/aarch64.c (aarch64_override_options_internal):
28223 Set flag_prefetch_loop_arrays according to tuning data.
28225 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
28227 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
28228 New tune structure.
28229 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
28230 [Unrelated to main purpose of the patch] Place the pointer field last
28231 to enable type checking errors when tune structure are wrongly merged.
28232 * config/aarch64/aarch64.c (generic_prefetch_tune,)
28233 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
28234 (thunderx2t99_prefetch_tune): New tune constants.
28235 (tune_params *_tunings): Update all tunings (no functional change).
28236 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
28237 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
28238 from tunings structures.
28240 2017-06-16 Jakub Jelinek <jakub@redhat.com>
28243 * ubsan.c (instrument_null): Add T argument, use it instead
28244 of computing it based on IS_LHS.
28245 (instrument_object_size): Likewise.
28246 (pass_ubsan::execute): Adjust instrument_null and
28247 instrument_object_size callers to pass gimple_get_lhs or
28248 gimple_assign_rhs1 result to it. Use instrument_null instead of
28249 calling get_base_address and instrument_mem_ref. Handle
28250 aggregate call arguments for object-size sanitization.
28252 2017-06-16 Yury Gribov <tetra2005@gmail.com>
28254 PR tree-optimization/81089
28255 * tree-vrp.c (is_masked_range_test): Validate operands of
28258 2017-06-15 Martin Sebor <msebor@redhat.com>
28261 * dumpfile.c (dump_register): Avoid calling memset to initialize
28262 a class with a default ctor.
28263 * gcc.c (struct compiler): Remove const qualification.
28264 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
28265 * hash-table.h: Ditto.
28266 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
28268 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
28269 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
28271 * params.h (struct param_info): Make struct members non-const.
28272 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
28273 with default initialization.
28274 * vec.h (vec_copy_construct, vec_default_construct): New helper
28276 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
28277 with vec_copy_construct.
28278 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
28279 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
28280 * doc/invoke.texi (-Wclass-memaccess): Document.
28282 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
28284 * emit-rtl.h (is_leaf): Update comment about local
28285 register allocator.
28287 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
28290 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
28291 for a variable to have a section before checking if the section has a
28293 Set section to.persistent if persistent attribute is set.
28294 Warn if .persistent attribute is used on an automatic variable.
28296 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
28298 PR rtl-optimization/80474
28299 * reorg.c (update_block): Do not ignore instructions in a delay slot.
28301 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
28303 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
28306 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
28308 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
28309 (casesi): Emit bounds checking as RTL.
28310 (casesi_internal_mips16_<mode>): Remove bounds checking.
28312 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
28314 * config/xtensa/xtensa.c (xtensa_option_override): Append
28315 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
28316 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
28317 xtensa_doloop_hooks): Define unconditionally.
28318 (xtensa_reorg_loops): Only call reorg_loops in the presence of
28320 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
28321 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
28322 for it in xtensa_option_override.
28323 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
28324 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
28326 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
28328 * doc/cppopts.texi: Document '-' special value to -MF.
28330 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
28332 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
28333 (cortex_a53_fconst): Likewise.
28334 (cortex_a53_fpmul): Likewise.
28335 (cortex_a53_f_load_64): Likewise.
28336 (cortex_a53_f_load_many): Likewise.
28337 (cortex_a53_advsimd_alu): Likewise.
28338 (cortex_a53_advsimd_alu_q): Likewise.
28339 (cortex_a53_advsimd_mul): Likewise.
28340 (cortex_a53_advsimd_mul_q): Likewise.
28341 (fpmac bypass): Add new bypass for fpmac-fpmac case.
28342 Add missing fmul, r2f_cvt and fconst cases.
28344 2017-06-14 Richard Biener <rguenther@suse.de>
28346 PR middle-end/81088
28347 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
28349 (fold_binary_loc): When associating do not treat pre-existing
28350 TREE_OVERFLOW on literal constants as a reason to allow
28351 TREE_OVERFLOW on associated literal constants.
28353 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28355 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
28356 (MASK_FEATURES): New macro.
28357 * config/sparc/sparc.c (sparc_option_override): Remove the special
28358 handling of -mfpu and generalize it to all MASK_FEATURES switches.
28360 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28362 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
28363 a division of 0 if non-call exceptions are enabled.
28365 2017-06-14 Andrew Pinski <apinski@cavium.com>
28366 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
28369 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
28370 Improve vector initialization code gen for only variable case.
28372 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
28374 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
28376 2017-06-14 Richard Biener <rguenther@suse.de>
28378 PR tree-optimization/81083
28379 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
28382 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28384 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
28385 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
28386 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
28387 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
28388 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
28389 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
28391 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28393 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
28394 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
28396 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28398 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
28400 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28402 * config/rs6000/t-rtems: Don't handle SPE.
28404 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28406 * config/rs6000/t-linux: Don't handle SPE.
28408 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28410 * config/rs6000/eabispe.h: Delete file.
28412 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28414 * config/rs6000/t-spe: Delete file.
28416 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28418 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
28419 (rs6000_legitimate_offset_address_p): Return false for anything in
28420 V2SImode or V2SFmode.
28422 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
28424 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
28425 except V2SF and V2SI. Rearrange the vector modes, and add comments.
28426 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
28428 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
28429 (rs6000_legitimate_offset_address_p): Ditto.
28430 (rs6000_emit_move): Ditto.
28431 (rs6000_init_builtins): Remove V4HI_type_node.
28433 2017-06-13 Martin Liska <mliska@suse.cz>
28436 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
28437 (gate_asan): Likewise.
28438 * asan.h (asan_no_sanitize_address_p): Remove the function.
28439 (sanitize_flags_p): New function.
28440 * builtins.def: Fix coding style.
28441 * common.opt: Use renamed enum value.
28442 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
28443 * doc/extend.texi: Document no_sanitize attribute.
28444 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
28445 to SANITIZE_UNDEFINED_NONDEFAULT.
28446 * gcc.c (sanitize_spec_function): Use the renamed enum value.
28447 * gimple-fold.c (optimize_atomic_compare_exchange_p):
28448 Use sanitize_flags_p.
28449 * gimplify.c (gimplify_function_tree): Likewise.
28450 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
28451 * opts.c (parse_no_sanitize_attribute): New function.
28452 (common_handle_option): Use renamed enum value.
28453 * opts.h (parse_no_sanitize_attribute): Declare.
28454 * tree.c (sanitize_flags_p): New function.
28455 * tree.h: Declared here.
28456 * tsan.c: Use sanitize_flags_p.
28457 * ubsan.c (ubsan_expand_null_ifn): Likewise.
28458 (instrument_mem_ref): Likewise.
28459 (instrument_bool_enum_load): Likewise.
28460 (do_ubsan_in_current_function): Remove the function.
28461 (pass_ubsan::execute): Use sanitize_flags_p.
28462 * ubsan.h: Remove do_ubsan_in_current_function
28463 * tree-cfg.c (print_no_sanitize_attr_value): New function.
28464 (dump_function_to_file): Use it here.
28466 2017-06-13 Martin Jambor <mjambor@suse.cz>
28468 PR tree-optimization/80803
28469 PR tree-optimization/81063
28470 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
28471 (propagate_subaccesses_across_link): Enqueue subtree whenever
28472 necessary instead of relying on the caller.
28474 2017-06-13 Martin Jambor <mjambor@suse.cz>
28476 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
28477 that have a first_link.
28478 (sort_and_splice_var_accesses): Do not check first_link before
28480 (subtree_mark_written_and_enqueue): Likewise.
28481 (propagate_all_subaccesses): Likewise and do not stop at first
28482 parent with a first_link.
28484 2017-06-13 Martin Jambor <mjambor@suse.cz>
28486 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
28489 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28491 * match.pd: New pattern.
28493 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28495 * tree-vrp.c (is_masked_range_test): New function.
28496 (register_edge_assert_for): Determine ranges for
28499 2017-06-13 Yury Gribov <tetra2005@gmail.com>
28501 PR tree-optimization/67328
28502 * fold-const.c (maskable_range_p): New function.
28503 (build_range_check): Generate bittests if possible.
28505 2017-06-13 Martin Liska <mliska@suse.cz>
28507 * gimple-pretty-print.c (dump_probability): Add new argument.
28508 (dump_edge_probability): Dump both probability and count.
28509 (dump_gimple_label): Likewise.
28510 (dump_gimple_bb_header): Likewise.
28512 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
28515 * config/avr/avr-devices.c: Fix indentation.
28516 * config/avr/gen-avr-mmcu-specs.c: Dito.
28518 2017-06-13 Richard Biener <rguenther@suse.de>
28520 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
28521 instead get vector type from stmt_info.
28522 (vectorizable_reduction): Adjust. Remove dead code.
28524 2017-06-13 Richard Biener <rguenther@suse.de>
28526 PR middle-end/81065
28527 * fold-const.c (extract_muldiv_1): Remove bogus distribution
28528 case of C * (x * C2 + C3).
28529 (fold_addr_of_array_ref_difference): Properly fold index difference.
28531 2017-06-12 David S. Miller <davem@davemloft.net>
28534 * config/sparc/sparc.md (return expander): Emit frame blockage if
28535 function uses alloca.
28537 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
28539 * combine.c (make_field_assignment): Check len rather than the mode
28540 precision when calling force_to_mode.
28542 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
28544 Support multilibs and devices that see flash in RAM address range.
28547 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
28548 (avr_mcu_t) <flash_pm_offset>: New field.
28549 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
28550 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
28551 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
28552 (AVR_TINY_PM_OFFSET): Remove macro.
28553 * config/avr/avr.opt (-mshort-calls): New option.
28554 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
28555 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
28556 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
28557 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
28558 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
28559 instead of avr_arch->have_jmp_call.
28560 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
28561 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
28562 avr_arch->flash_pm_offset to define.
28563 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
28564 new field flash_pm_offset. Add entry for avrxmega3.
28565 (avr_texinfo): Add entry for avrxmega3.
28566 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
28567 attiny212, attiny214,
28568 attiny412, attiny414, attiny416, attiny417,
28569 attiny814, attiny816, attiny817,
28570 attiny1614, attiny1616, attiny1617,
28571 attiny3214, attiny3216, attiny3217.
28572 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
28573 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
28574 (avr_print_operand_address) [AVR_TINY]: Same.
28575 (avr_asm_init_sections) <readonly_data_section>: Only patch
28576 callback if avr_arch->flash_pm_offset = 0.
28577 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
28578 for rodata if avr_arch->flash_pm_offset != 0.
28579 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
28580 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
28581 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
28582 (m_options): Append opt_rcall.
28583 (m_dirnames): Append dir_rcall.
28584 * config/avr/t-multilib: Regenerate.
28586 * configure.ac [target=avr]: Check whether avrxmega3 default
28587 linker description file works as needed.
28588 * configure: Regenerate.
28589 * doc/avr-mmcu.texi: Regenerate.
28590 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
28591 <__AVR_ARCH__>: Document avrxmega3 and 103.
28592 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
28593 <__AVR_SHORT_CALLS__>: Document it.
28594 <__AVR_PM_BASE_ADDRESS__>: Document it.
28595 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
28596 (AVR Variable Attributes) <progmem>: Document this is
28597 not needed for avrxmega3.
28598 (AVR Named Address Spaces) <__flash>: Dito.
28600 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
28602 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
28604 2017-06-12 Doug Rupp <rupp@adacore.com>
28606 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
28607 Append vxworks-stdint.h to the tm_file list.
28608 * config/vxworks-stdint.h: New file.
28610 2017-06-12 Martin Liska <mliska@suse.cz>
28612 PR tree-optimization/81041
28613 * tree-profile.c (gimple_gen_ic_func_profiler):
28614 Create an extra BB in profile-generate
28615 (gimple_gen_time_profiler): Likewise.
28617 2017-06-12 Jakub Jelinek <jakub@redhat.com>
28619 PR tree-optimization/81003
28620 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
28621 (update_range_test): Use it instead of force_gimple_operand_gsi.
28623 2017-06-12 Richard Biener <rguenther@suse.de>
28625 PR tree-optimization/81053
28626 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
28627 with backedge value not defined in loop. Simplify def stmt
28630 2017-06-11 Tom de Vries <tom@codesourcery.com>
28633 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
28635 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
28636 nvptx_cannot_force_const_mem.
28638 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28640 * opts.c (finish_options): Move test for flag_split_stack after
28641 it has been initialized.
28643 2017-06-11 Jason Merrill <jason@redhat.com>
28645 * tree.h (id_equal): New.
28646 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
28647 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
28648 instead of strcmp of IDENTIFIER_POINTER.
28650 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28652 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
28653 (mark_all_inlined_calls_cdtor): Fix formating.
28654 (inline_transform): Rescale profile before inlining.
28656 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28658 * cgraph.h (cgraph_edge::clone): Update prototype.
28659 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
28660 (cgraph_node::create_clone): Update.
28661 (cgraph_node::create_version_clone): Update.
28662 * tree-inline.c (copy_bb): Update.
28663 (expand_call_inline): Update.
28665 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28667 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
28668 factored out from ...
28669 (rs6000_emit_prologue): ... here.
28671 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
28673 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
28674 factored out from ...
28675 (rs6000_emit_prologue): ... here.
28677 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28679 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
28681 (handle_missing_profiles): Fix computation of tp_first_run.
28682 (counts_to_freqs): Do not touch freqs when count is 0.
28684 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
28686 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
28689 2017-06-10 Tom de Vries <tom@codesourcery.com>
28691 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28692 attributes): Document signal effective target.
28694 2017-06-10 Tom de Vries <tom@codesourcery.com>
28696 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28697 Document effective target stack_size.
28699 2017-06-09 David Malcolm <dmalcolm@redhat.com>
28701 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
28702 to the edit_context if they can be auto-applied.
28704 2017-06-9 Ian Lance Taylor <iant@golang.org>
28706 * opts.c (finish_options): If -fsplit-stack, disable implicit
28707 -forder-blocks-and-partition.
28708 * doc/invoke.texi (Optimize Options): Document that when using
28709 -fsplit-stack -forder-blocks-and-partition is not implicitly
28712 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28714 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
28715 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
28716 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
28717 * builtins.def (abort, trap, unreachable): Declare cold.
28718 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
28719 * tree-core.h (ECF_COLD): New.
28720 * tree.c (set_call_expr_flags): Handle ECF_COLD.
28721 (build_common_builtin_nodes): Mark unreachable and abort as cold.
28723 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28725 * predict.c (unlikely_executed_stmt_p): Cleanup.
28727 2017-06-09 Richard Biener <rguenther@suse.de>
28729 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
28730 model if the ref is always written to.
28732 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28734 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
28736 2017-06-09 Tamar Christina <tamar.christina@arm.com>
28738 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
28741 2017-06-09 Tom de Vries <tom@codesourcery.com>
28744 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
28745 "target cannot support label values" when encountering LABEL_REF.
28747 2017-06-09 Martin Liska <mliska@suse.cz>
28749 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
28750 (gimple_gen_ic_func_profiler): Emit direct comparison
28751 of __gcov_indirect_call_callee with NULL.
28752 (gimple_gen_time_profiler): Change probability from
28753 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
28755 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28757 * profile.c (edge_gcov_counts): Turn to pointer.
28758 (compute_branch_probabilities, compute_branch_probabilities): Update.
28759 (branch_prob): Do not clear edge_gcov_count.
28760 * profile.h (edge_gcov_counts): Turn to pointer.
28761 (edge_gcov_count): Update.
28763 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
28765 * gimple.h (gimple_check_failed): Mark cold.
28767 2017-06-09 Richard Biener <rguenther@suse.de>
28769 PR tree-optimization/66623
28770 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
28771 refactor check_reduction into two parts, properly computing
28772 whether we have to check reduction validity for outer loop
28775 2017-06-09 Richard Biener <rguenther@suse.de>
28777 PR tree-optimization/79483
28778 * graphite-scop-detection.c (order): New global.
28779 (get_order): Compute bb to order mapping that satisfies code
28780 generation constraints.
28781 (cmp_pbbs): New helper.
28782 (build_scops): Start domwalk at entry block, sort generated
28785 2017-06-09 Richard Biener <rguenther@suse.de>
28787 PR middle-end/81007
28788 * ipa-polymorphic-call.c
28789 (ipa_polymorphic_call_context::restrict_to_inner_class):
28790 Skip FIELD_DECLs with error_mark_node type.
28791 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
28794 2017-06-09 Martin Liska <mliska@suse.cz>
28796 * predict.c (struct branch_predictor): New struct.
28797 (test_prediction_value_range): New test.
28798 (predict_c_tests): New function.
28799 * selftest-run-tests.c (selftest::run_tests): Run the function.
28800 * selftest.h: Declare new tests.
28802 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
28805 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
28806 gen_add3_insn did not fail.
28807 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
28808 r0, construct that number in a temporary reg and add that reg to r0.
28809 If asked to put the result in r0 as well, fail.
28811 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
28813 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
28814 for early expansion of vec_eqv.
28816 2017-06-08 Jakub Jelinek <jakub@redhat.com>
28818 PR middle-end/81005
28819 * ubsan.c (instrument_null): Avoid pointless code temporary.
28820 (pass_ubsan::execute): Instrument aggregate arguments of calls.
28822 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
28826 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
28829 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
28832 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28834 * predict.c (unlikely_executed_edge_p): Move ahead.
28835 (probably_never_executed_edge_p): Use it.
28837 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28839 PR middle-end/79988
28840 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
28841 gimple_call_builtin_p call.
28843 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28845 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
28846 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
28847 rtl_check_failed_type2, rtl_check_failed_code1,
28848 rtl_check_failed_code2, rtl_check_failed_code_mode,
28849 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
28850 rtvec_check_failed_bounds, rtl_check_failed_flag,
28851 _fatal_insn_not_found, _fatal_insn): Likewise.
28852 * tree.h (tree_contains_struct_check_failed,
28853 tree_check_failed, tree_not_check_failed,
28854 tree_class_check_failed, tree_range_check_failed,
28855 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
28856 tree_vec_elt_check_failed, phi_node_elt_check_failed,
28857 tree_operand_check_failed, omp_clause_check_failed,
28858 omp_clause_operand_check_failed, omp_clause_range_check_failed):
28861 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28863 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
28864 flag_branch_probabilities.
28865 * ipa-inline.c (edge_badness): Likewise.
28866 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
28867 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
28868 * predict.c (maybe_hot_frequency_p): Likewise.
28869 (probably_never_executed): Likewise.
28870 * sched-ebb.c (schedule_ebbs): Likewise.
28871 * sched-rgn.c (find_single_block_region): Likewise.
28872 * tracer.c (tail_duplicate): Likewise.
28874 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28876 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
28877 longer requires x_flag_profile_use.
28879 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28881 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
28882 instead of flag_reorder_blocks_and_partition.
28883 * dbxout.c (dbxout_function_end): Likewise.
28884 * dwarf2out.c (gen_subprogram_die): Likewise.
28885 * haifa-sched.c (sched_create_recovery_edges): Likewise.
28886 * hw-doloop.c (reorg_loops): Likewise.
28887 * varasm.c (assemble_start_function,
28888 assemble_end_function): Likewise.
28889 (decide_function_section): Do not check for
28890 flag_reorder_blocks_and_partition.
28892 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
28894 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
28896 (chkp_get_hard_register_fake_addr_expr): Ditto.
28897 (chkp_build_addr_expr): Add check for hard reg case.
28898 (chkp_parse_array_and_component_ref): Ditto.
28899 (chkp_find_bounds_1): Ditto.
28900 (chkp_process_stmt): Don't generate bounds store for
28903 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28905 * predict.c (maybe_hot_bb_p): Do not check profile status.
28906 (maybe_hot_edge_p): Likewise.
28907 (probably_never_executed): Check for zero counts even if profile
28909 (unlikely_executed_edge_p): New function.
28910 (unlikely_executed_stmt_p): New function.
28911 (unlikely_executed_bb_p): New function.
28912 (set_even_probabilities): Use unlikely predicates.
28913 (combine_predictions_for_bb): Likewise.
28914 (predict_paths_for_bb): Likewise.
28915 (predict_paths_leading_to_edge): Likewise.
28916 (determine_unlikely_bbs): New function.
28917 (estimate_bb_frequencies): Use it.
28918 (compute_function_frequency): Use zero counts even if profile is
28920 * profile-count.h: Fix typo.
28922 2017-08-08 Julia Koval <julia.koval@intel.com>
28924 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
28925 _mm512_mask_cvtsepi16_storeu_epi8,
28926 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
28927 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
28928 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
28929 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
28930 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
28931 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
28932 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
28933 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
28934 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
28935 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
28936 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
28937 __builtin_ia32_pmovuswb256mem_mask,
28938 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
28939 __builtin_ia32_pmovwb512mem_mask): New builtins.
28941 2017-08-08 Julia Koval <julia.koval@intel.com>
28943 PR target/73350,80862
28944 * config/i386/subst.md (round): Fix round pattern.
28945 * config/i386/i386.c (ix86_erase_embedded_rounding):
28946 Fix erasing rounding for the fixed pattern.
28948 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
28950 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
28952 2017-06-08 Martin Liska <mliska@suse.cz>
28954 PR gcov-profile/80911
28955 * gcov.c (block_info::block_info): New constructor.
28957 2017-06-07 Carl Love <cel@us.ibm.com>
28959 * config/rs6000/rs6000-c: The return type of the following
28960 built-in functions was implemented as int not long long. Fix sign
28961 of return value for the unsigned version of vec_mulo and vec_mule.
28962 vector unsigned long long vec_bperm (vector unsigned long long,
28963 vector unsigned char)
28964 vector signed long long vec_mule (vector signed int,
28966 vector unsigned long long vec_mule (vector unsigned int,
28967 vector unsigned int)
28968 vector signed long long vec_mulo (vector signed int,
28970 vector unsigned long long vec_mulo (vector unsigned int,
28971 vector unsigned int)
28972 * doc/extend.texi: Fix the documentation for the built-in
28975 2017-06-07 Carl Love <cel@us.ibm.com>
28978 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
28981 2017-06-07 Carl Love <cel@us.ibm.com>
28983 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
28984 support, Generate doublehv for signed int/float for BE case only.
28986 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
28988 * doc/invoke.texi (mcx16): Rewrite.
28990 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28992 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
28993 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
28994 *mov<mode>_softfloat, and an anonymous splitter): Use
28995 nonimmediate_operand instead of rs6000_nonimmediate_operand.
28997 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
28999 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
29001 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
29002 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
29003 (rs6000_debug_reg_global): Adjust.
29004 (rs6000_init_hard_regno_mode_ok): Adjust.
29005 (rs6000_dbx_register_number): Adjust.
29006 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
29007 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
29008 Remove SPE_ACC and SPEFSCR.
29009 (REG_ALLOC_ORDER): Ditto.
29010 (FRAME_POINTER_REGNUM): Change to 111.
29011 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
29012 (REG_CLASS_NAMES): Ditto.
29013 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
29014 (REGISTER_NAMES): Ditto.
29015 (ADDITIONAL_REG_NAMES): Ditto.
29016 (rs6000_reg_names): Ditto.
29017 * config/rs6000/rs6000.md: Renumber some register number
29020 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29022 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
29024 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
29025 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
29027 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
29028 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
29029 Delete the SPE high registers.
29030 (REG_ALLOC_ORDER): Ditto.
29031 (enum reg_class): Remove SPE_HIGH_REGS.
29032 (REG_CLASS_NAMES): Ditto.
29033 (REG_CLASS_CONTENTS): Delete the SPE high registers.
29034 (REGISTER_NAMES): Ditto.
29035 (rs6000_reg_names): Ditto.
29036 * doc/tm.texi.in: Remove SPE as example.
29037 * doc/tm.texi: Regenerate.
29039 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29041 * config/rs6000/8540.md (ppc8540_brinc): Delete.
29042 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
29043 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
29044 * config/rs6000/rs6000.md (type): Remove "brinc".
29046 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29048 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
29049 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
29050 * config/rs6000/linuxspe.h: Delete file.
29051 * config/rs6000/rs6000.md: Don't include spe.md.
29052 * config/rs6000/spe.h: Delete file.
29053 * config/rs6000/spe.md: Delete file.
29054 * config/rs6000/t-rs6000: Remove spe.md.
29056 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29058 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
29059 (reg_or_none500mem_operand): Delete.
29060 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
29061 instead of reg_or_none500mem_operand.
29063 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29065 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
29066 handling of SPE flags.
29067 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
29069 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29071 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
29073 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
29074 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
29075 paired_divv2sf3): Similar.
29076 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
29077 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
29078 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
29080 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
29081 Rename the paired_* instruction patterns.
29082 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
29084 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
29085 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
29086 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
29087 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
29088 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
29089 PAIRED_VECTOR_MODE.
29090 (struct machine_function): Delete field spe_insn_chain_scanned_p.
29091 (spe_func_has_64bit_regs_p): Delete.
29092 (spe_expand_predicate_builtin): Delete.
29093 (spe_expand_evsel_builtin): Delete.
29094 (TARGET_DWARF_REGISTER_SPAN): Do not define.
29095 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
29096 (invalid_e500_subreg): Delete.
29097 (rs6000_legitimize_address): Always force_reg op2 as well, for
29098 paired single memory accesses.
29099 (rs6000_member_type_forces_blk): Delete.
29100 (rs6000_spe_function_arg): Delete.
29101 (rs6000_expand_unop_builtin): Delete SPE handling.
29102 (rs6000_expand_binop_builtin): Ditto.
29103 (spe_expand_stv_builtin): Delete.
29104 (bdesc_2arg_spe): Delete.
29105 (spe_expand_builtin): Delete.
29106 (spe_expand_predicate_builtin): Delete.
29107 (spe_expand_evsel_builtin): Delete.
29108 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
29109 (spe_init_builtins): Delete.
29110 (spe_func_has_64bit_regs_p): Delete.
29111 (savres_routine_name): Delete "info" parameter. Adjust callers.
29112 (rs6000_emit_stack_reset): Ditto.
29113 (rs6000_dwarf_register_span): Delete.
29114 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
29115 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
29116 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
29118 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
29120 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
29121 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
29122 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
29123 mulv2sf3, divv2sf3): Delete expanders.
29125 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29127 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
29129 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29131 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
29132 * config/rs6000/rs6000.c: Ditto.
29134 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29136 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
29137 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
29138 comparison_operator.
29140 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29142 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
29143 * config/rs6000/rs6000.opt: Ditto.
29144 * config/rs6000/t-rtems: Ditto.
29146 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29148 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
29149 TARGET_E500_SINGLE by 0, simplify.
29150 * config/rs6000/rs6000.c: Ditto.
29151 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
29152 (spe_build_register_parallel): Delete.
29153 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
29154 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
29155 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
29156 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
29157 (E500_CONVERT): Delete.
29158 * config/rs6000/spe.md: Remove many patterns and all define_constants.
29160 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
29162 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
29163 * config/rs6000/dfp.md: Ditto.
29164 (negdd2, *negdd2_fpr): Merge.
29165 (absdd2, *absdd2_fpr): Merge.
29166 (negtd2, *negtd2_fpr): Merge.
29167 (abstd2, *abstd2_fpr): Merge.
29168 * config/rs6000/e500.h: Delete file.
29169 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
29170 TARGET_FPRS by 1 and simplify.
29171 * config/rs6000/rs6000-c.c: Ditto.
29172 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
29173 TARGET_DF_SPE by 0.
29174 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
29176 * config/rs6000/rs6000.md: Ditto.
29177 (floatdidf2, *floatdidf2_fpr): Merge.
29178 (move_from_CR_gt_bit): Delete.
29179 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
29180 (E500_CR_IOR_COMPARE): Delete.
29181 (All patterns that require !TARGET_FPRS): Delete.
29182 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
29184 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29186 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
29188 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29190 * graphds.c (add_edge): Intitialize edge's attached data.
29191 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
29192 pointer parameter. Call pointed function on each edge during
29193 graph traversing. Skip traversing the edge when the function
29195 (graphds_dfs, graphds_scc): Ditto.
29196 (for_each_edge): New parameter. Pass the new parameter to callback
29198 * graphds.h (skip_edge_callback): New function pointer type.
29199 (graphds_dfs, graphds_scc): New function pointer parameter.
29200 (graphds_edge_callback, for_each_edge): New parameter.
29202 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29204 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
29205 out code checking if runtime alias check is possible to below ...
29206 Call the new function.
29207 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
29208 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
29210 2017-06-07 Marek Polacek <polacek@redhat.com>
29213 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
29214 TYPE_OVERFLOW_WRAPS check.
29216 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29218 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
29219 if versioning is required.
29220 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
29221 peeling with the check for versioning.
29223 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29225 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
29226 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
29227 Set true to new parameter if new ssa variable is defined.
29228 (vect_gen_vector_loop_niters): Refactor. Set range information
29229 for the new vector loop bound variable.
29230 (vect_do_peeling): Ditto.
29232 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29234 * tree-affine.c (ssa.h): Include header file.
29235 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
29236 has wrapping overflow behavior.
29238 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29240 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
29242 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29244 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
29245 (tree_to_aff_combination): ... here.
29247 2017-06-07 Bin Cheng <bin.cheng@arm.com>
29249 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
29250 reg_pressure model function.
29251 (ivopts_global_cost_for_size): Delete.
29252 (determine_set_costs, iv_ca_recount_cost): Call new model function
29253 ivopts_estimate_reg_pressure.
29255 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29257 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
29258 expensive than udiv. Remove floating point cases from mod.
29260 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29262 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
29263 Increase idiv cost.
29265 2017-06-07 Tamar Christina <tamar.christina@arm.com>
29267 * config/aarch64/aarch64.md
29268 (copysignsf3): Fix mask generation.
29270 2017-06-07 Jakub Jelinek <jakub@redhat.com>
29272 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
29274 (class dump_manager): Add register_dumps method.
29275 * dumpfile.c: Include langhooks.h.
29276 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
29277 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
29278 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
29279 (dump_manager::dump_register): Start with 512 entries instead of 32.
29280 (dump_manager::register_dumps): New method.
29281 * toplev.c (general_init): Instead of invoking register_dumps
29282 langhook, invoke register_dumps method on the dump manager.
29283 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
29286 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
29288 * doc/md.texi: Clarify the restrictions on a define_insn condition.
29289 Say that # requires an associated define_split to exist, and that
29290 the define_split must be suitable for use after register allocation.
29292 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29294 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
29295 (compute_outgoing_frequencies): Also initialize zero counts.
29296 (find_many_sub_basic_blocks): Do not produce uninitialized profile
29297 around loops; preserve more of profile when nothing changes.
29299 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
29301 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
29303 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
29304 * config/arm/arm-cpu-cdata.h: Regenerate.
29305 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
29306 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
29307 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
29308 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
29309 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
29311 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
29313 * config/arm/t-rmprofile: Likewise.
29314 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
29316 2017-06-06 David S. Miller <davem@davemloft.net>
29319 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
29320 blockage if function uses alloca.
29322 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29324 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
29325 New "uid" fields to hold pretty-print IDs of group and ref.
29326 Memory references are now identified as <group_id>:<ref_id>
29327 instead of using [random] addresses.
29328 (dump_mem_details): Simplify, no functional change.
29329 (dump_mem_ref): Simplify and make output more concise.
29330 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
29331 (find_or_create_group): Initialize group uid.
29332 (record_ref): Initialize ref uid. Improve debug output.
29333 (prune_group_by_reuse, should_issue_prefetch_p,)
29334 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
29335 (mark_nontemporal_store, determine_loop_nest_reuse):
29336 Improve debug output.
29338 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
29340 * dbgcnt.def (prefetch): New debug counter.
29341 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
29342 (schedule_prefetches): Stop issueing prefetches if debug counter
29345 2017-06-06 Tom de Vries <tom@codesourcery.com>
29347 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
29348 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
29350 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29352 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
29353 Use aarch64_reg_or_zero predicate for operand 4.
29354 (aarch64_compare_and_swap<mode> define_insn_and_split):
29355 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
29356 (aarch64_store_exclusive<mode>): Likewise for operand 2.
29358 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
29360 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
29361 (arm_compute_save_core_reg_mask): This.
29362 (thumb1_compute_save_reg_mask): Rename into ...
29363 (thumb1_compute_save_core_reg_mask): This.
29364 (arm_compute_save_reg0_reg12_mask): Adapt comment.
29365 (arm_compute_frame_layout): Likewise.
29367 2017-06-06 Richard Biener <rguenther@suse.de>
29369 PR tree-optimization/80974
29370 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
29371 keep or clear leaders SSA info.
29373 2017-06-06 Tom de Vries <tom@codesourcery.com>
29375 * config/nvptx/nvptx.c (split_mode_p): New function.
29376 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
29378 2017-06-06 Tom de Vries <tom@codesourcery.com>
29380 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
29382 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29385 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
29388 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29390 * shrink-wrap.c (handle_simple_exit): Update profile.
29391 (try_shrink_wrapping): Upate profile.
29393 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29395 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
29396 (tree_guess_outgoing_edge_probabilities): New.
29397 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
29398 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
29400 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29402 * ipa-split.c (split_function): Initialize return bb profile.
29404 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
29406 * profile.c (compute_branch_probabilities): Also initialize
29407 EXIT_BLOCK profile.
29409 2017-06-06 Richard Biener <rguenther@suse.de>
29411 PR tree-optimization/80928
29412 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
29413 (vect_analyze_loop_operations): Properly guard analysis for
29415 (vect_transform_loop): Likewise.
29416 (vect_analyze_loop_2): Also reset SLP type on PHIs.
29417 (vect_model_induction_cost): Do not cost for pure SLP.
29418 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
29419 of induction in inner loop vectorization.
29420 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
29421 (vect_get_and_check_slp_defs): Handle vect_induction_def.
29422 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
29424 (vect_analyze_slp_cost_1): Cost induction.
29425 (vect_detect_hybrid_slp_stmts): Handle PHIs.
29426 (vect_get_slp_vect_defs): Likewise.
29427 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
29428 (vect_transform_stmt): Handle SLP reductions.
29429 * tree-vectorizer.h (vectorizable_induction): Adjust.
29431 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29433 * config/rs6000/rs6000.c (make_resolver_func): Update
29434 init_lowered_empty_function call.
29436 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
29438 * doc/invoke.texi: Document the -fprofile-abs-path option.
29439 * common.opt (fprofile-abs-path): New option.
29440 * gcov-io.h (gcov_write_filename): Declare.
29441 * gcov-io.c (gcov_write_filename): New function.
29442 * coverage.c (coverage_begin_function): Use gcov_write_filename.
29443 * profile.c (output_location): Likewise.
29445 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29447 * shring-wrap.c: Revert accidental commit.
29449 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
29451 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
29453 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29455 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
29457 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
29458 profile in callgraph edge.
29459 * profile-count.h (apply_probability): If THIS is 0, then result is 0
29460 (apply_scale): Likewise.
29461 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
29462 Also scale profile when inlining function with zero profile.
29463 (initialize_cfun): Update exit block profile even when it is zero.
29464 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
29465 when profile is read.
29467 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
29469 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
29470 (CLONE_*): New constants to define the processors we can generate
29471 code for with the target_clone attribute.
29472 (rs6000_clone_map): New array to identify which clone processors
29473 the current program is running on.
29474 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
29475 target_clone attribute.
29476 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
29477 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
29478 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
29479 (cpu_expand_builtin): Add support for target_clone attribute.
29480 (rs6000_valid_attribute_p): Allow "default" attribute.
29481 (get_decl_name): New debug function to simplify printing the
29482 current function name in debugging statements.
29483 (rs6000_clone_priority): New functions to support the target_clone
29484 attribute, and be able to generate code to switch between ISA 2.05
29485 through ISA 3.0 (power6 through power9).
29486 (rs6000_compare_version_priority): Likewise.
29487 (rs6000_get_function_versions_dispatcher): Likewise.
29488 (make_resolver_func): Likewise.
29489 (add_condition_to_bb): Likewise.
29490 (dispatch_function_versions): Likewise.
29491 (rs6000_generate_version_dispatcher_body): Likewise.
29492 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
29493 (fusion_gpr_load_p): Fix a spacing issue.
29494 * doc/extend.texi (Common Function Attributes): Document that the
29495 PowerPC supports the target_clone attribute.
29497 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
29499 * config/arm/arm.h: explain F symbol found in description of ARM
29500 register allocation in its legend.
29502 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29504 * config/mips/frame-header-opt.c: Include profile-count.h.
29505 * config/riscv/riscv.c: Include profile-count.h
29507 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
29509 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
29511 (sm_set_flag_if_changed): Add bbs field.
29512 (execute_sm_if_changed_flag_set): Pass BBS.
29513 (execute_sm): Update.
29515 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29517 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
29520 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29522 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
29523 (peephole2): New peephole2 to emit the above.
29524 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
29526 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29528 * config/aarch64/aarch64.c (define_peephole2 above
29529 *sub_<shift>_<mode>): New peephole.
29531 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
29533 * config/i386/i386.c (make_resolver_func): Update.
29534 * Makefile.in: Add profile-count.h and profile-count.o
29535 * auto-profile.c (afdo_indirect_call): Update to new API.
29536 (afdo_set_bb_count): Update.
29537 (afdo_propagate_edge): Update.
29538 (afdo_propagate_circuit): Update.
29539 (afdo_calculate_branch_prob): Update.
29540 (afdo_annotate_cfg): Update.
29541 * basic-block.h: Include profile-count.h
29542 (struct edge_def): Turn count to profile_count.
29543 (struct basic_block_def): Likewie.
29544 (REG_BR_PROB_BASE): Move to profile-count.h
29545 (RDIV): Move to profile-count.h
29546 * bb-reorder.c (max_entry_count): Turn to profile_count.
29547 (find_traces): Update.
29548 (rotate_loop):Update.
29549 (connect_traces):Update.
29550 (sanitize_hot_paths):Update.
29551 * bt-load.c (migrate_btr_defs): Update.
29552 * cfg.c (RDIV): Remove.
29553 (init_flow): Use alloc_block.
29554 (alloc_block): Uninitialize count.
29555 (unchecked_make_edge): Uninitialize count.
29556 (check_bb_profile): Update.
29557 (dump_edge_info): Update.
29558 (dump_bb_info): Update.
29559 (update_bb_profile_for_threading): Update.
29560 (scale_bbs_frequencies_int): Update.
29561 (scale_bbs_frequencies_gcov_type): Update.
29562 (scale_bbs_frequencies_profile_count): New.
29563 * cfg.h (update_bb_profile_for_threading): Update.
29564 (scale_bbs_frequencies_profile_count): Declare.
29565 * cfgbuild.c (compute_outgoing_frequencies): Update.
29566 (find_many_sub_basic_blocks): Update.
29567 * cfgcleanup.c (try_forward_edges): Update.
29568 (try_crossjump_to_edge): Update.
29569 * cfgexpand.c (expand_gimple_tailcall): Update.
29570 (construct_exit_block): Update.
29571 * cfghooks.c (verify_flow_info): Update.
29572 (dump_bb_for_graph): Update.
29573 (split_edge): Update.
29574 (make_forwarder_block): Update.
29575 (duplicate_block): Update.
29576 (account_profile_record): Update.
29577 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
29578 (get_estimated_loop_iterations): Update.
29579 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
29580 (single_likely_exit): Update.
29581 * cfgloopmanip.c (scale_loop_profile): Update.
29583 (set_zero_probability): Update.
29584 (lv_adjust_loop_entry_edge): Update.
29585 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
29586 (purge_dead_edges): Update.
29587 (rtl_account_profile_record): Update.
29588 * cgraph.c (cgraph_node::create): Uninitialize count.
29589 (symbol_table::create_edge): Uninitialize count.
29590 (cgraph_update_edges_for_call_stmt_node): Update.
29591 (cgraph_edge::dump_edge_flags): Update.
29592 (cgraph_node::dump): Update.
29593 (cgraph_edge::maybe_hot_p): Update.
29594 * cgraph.h: Include profile-count.h
29595 (create_clone), create_edge, create_indirect_edge): Update.
29596 (cgraph_node): Turn count to profile_count.
29597 (cgraph_edge0: Likewise.
29598 (make_speculative, clone): Update.
29599 (create_edge): Update.
29600 (init_lowered_empty_function): Update.
29601 * cgraphclones.c (cgraph_edge::clone): Update.
29602 (duplicate_thunk_for_node): Update.
29603 (cgraph_node::create_clone): Update.
29604 * cgraphunit.c (cgraph_node::analyze): Update.
29605 (cgraph_node::expand_thunk): Update.
29606 * final.c (dump_basic_block_info): Update.
29607 * gimple-streamer-in.c (input_bb): Update.
29608 * gimple-streamer-out.c (output_bb): Update.
29609 * graphite.c (print_global_statistics): Update.
29610 (print_graphite_scop_statistics): Update.
29611 * hsa-brig.c: Include basic-block.h.
29612 * hsa-dump.c: Include basic-block.h.
29613 * hsa-gen.c (T sum_slice): Update.
29614 (convert_switch_statements):Update.
29615 * hsa-regalloc.c: Include basic-block.h.
29616 * ipa-chkp.c (chkp_produce_thunks): Update.
29617 * ipa-cp.c (struct caller_statistics): Update.
29618 (init_caller_stats): Update.
29619 (gather_caller_stats): Update.
29620 (ipcp_cloning_candidate_p): Update.
29621 (good_cloning_opportunity_p): Update.
29622 (get_info_about_necessary_edges): Update.
29623 (dump_profile_updates): Update.
29624 (update_profiling_info): Update.
29625 (update_specialized_profile): Update.
29626 (perhaps_add_new_callers): Update.
29627 (decide_about_value): Update.
29628 (ipa_cp_c_finalize): Update.
29629 * ipa-devirt.c (struct odr_type_warn_count): Update.
29630 (struct decl_warn_count): Update.
29631 (struct final_warning_record): Update.
29632 (possible_polymorphic_call_targets): Update.
29633 (ipa_devirt): Update.
29634 * ipa-fnsummary.c (redirect_to_unreachable): Update.
29635 * ipa-icf.c (sem_function::merge): Update.
29636 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
29637 * ipa-inline.c (compute_uninlined_call_time): Update.
29638 (compute_inlined_call_time): Update.
29639 (want_inline_small_function_p): Update.
29640 (want_inline_self_recursive_call_p): Update.
29641 (edge_badness): Update.
29642 (lookup_recursive_calls): Update.
29643 (recursive_inlining): Update.
29644 (inline_small_functions): Update.
29645 (dump_overall_stats): Update.
29646 (dump_inline_stats): Update.
29647 * ipa-profile.c (ipa_profile_generate_summary): Update.
29648 (ipa_propagate_frequency): Update.
29649 (ipa_profile): Update.
29650 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
29651 * ipa-utils.c (ipa_merge_profiles): Update.
29652 * loop-doloop.c (doloop_modify): Update.
29653 * loop-unroll.c (report_unroll): Update.
29654 (unroll_loop_runtime_iterations): Update.
29655 * lto-cgraph.c (lto_output_edge): Update.
29656 (lto_output_node): Update.
29657 (input_node): Update.
29658 (input_edge): Update.
29659 (merge_profile_summaries): Update.
29660 * lto-streamer-in.c (input_cfg): Update.
29661 * lto-streamer-out.c (output_cfg): Update.
29662 * mcf.c (create_fixup_graph): Update.
29663 (adjust_cfg_counts): Update.
29664 (sum_edge_counts): Update.
29665 * modulo-sched.c (sms_schedule): Update.
29666 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
29667 * predict.c (maybe_hot_count_p): Update.
29668 (probably_never_executed): Update.
29669 (dump_prediction): Update.
29670 (combine_predictions_for_bb): Update.
29671 (propagate_freq): Update.
29672 (handle_missing_profiles): Update.
29673 (counts_to_freqs): Update.
29674 (rebuild_frequencies): Update.
29675 (force_edge_cold): Update.
29676 * predict.h: Include profile-count.h
29677 (maybe_hot_count_p, counts_to_freqs): UPdate.
29678 * print-rtl-function.c: Do not include cfg.h
29679 * print-rtl.c: Include basic-block.h
29680 * profile-count.c: New file.
29681 * profile-count.h: New file.
29682 * profile.c (is_edge_inconsistent): Update.
29683 (correct_negative_edge_counts): Update.
29684 (is_inconsistent): Update.
29685 (set_bb_counts): Update.
29686 (read_profile_edge_counts): Update.
29687 (compute_frequency_overlap): Update.
29688 (compute_branch_probabilities): Update; Initialize and deinitialize
29690 (branch_prob): Update.
29691 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
29692 (edge_gcov_count): New.
29693 (bb_gcov_count): New.
29694 * shrink-wrap.c (try_shrink_wrapping): Update.
29695 * tracer.c (better_p): Update.
29696 * trans-mem.c (expand_transaction): Update.
29697 (ipa_tm_insert_irr_call): Update.
29698 (ipa_tm_insert_gettmclone_call): Update.
29699 * tree-call-cdce.c: Update.
29700 * tree-cfg.c (gimple_duplicate_sese_region): Update.
29701 (gimple_duplicate_sese_tail): Update.
29702 (gimple_account_profile_record): Update.
29703 (execute_fixup_cfg): Update.
29704 * tree-inline.c (copy_bb): Update.
29705 (copy_edges_for_bb): Update.
29706 (initialize_cfun): Update.
29707 (freqs_to_counts): Update.
29708 (copy_cfg_body): Update.
29709 (expand_call_inline): Update.
29710 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
29711 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
29712 (try_unroll_loop_completely): Update.
29713 (try_peel_loop): Update.
29714 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
29715 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
29716 * tree-ssa-loop-split.c (connect_loops): Update.
29717 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
29718 * tree-ssa-reassoc.c (branch_fixup): Update.
29719 * tree-ssa-tail-merge.c (replace_block_by): Update.
29720 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
29721 (compute_path_counts): Update.
29722 (update_profile): Update.
29723 (recompute_probabilities): Update.
29724 (update_joiner_offpath_counts): Update.
29725 (estimated_freqs_path): Update.
29726 (freqs_to_counts_path): Update.
29727 (clear_counts_path): Update.
29728 (ssa_fix_duplicate_block_edges): Update.
29729 (duplicate_thread_path): Update.
29730 * tree-switch-conversion.c (case_bit_test_cmp): Update.
29731 (struct switch_conv_info): Update.
29732 * tree-tailcall.c (decrease_profile): Update.
29733 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
29734 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
29735 * value-prof.c (check_counter): Update.
29736 (gimple_divmod_fixed_value): Update.
29737 (gimple_mod_pow2): Update.
29738 (gimple_mod_subtract): Update.
29739 (gimple_ic_transform): Update.
29740 (gimple_stringop_fixed_value): Update.
29741 * value-prof.h (gimple_ic): Update.
29743 2017-06-02 Carl Love <cel@us.ibm.com>
29745 * config/rs6000/rs6000-c: Add support for built-in functions
29746 vector double vec_doublee (vector signed int);
29747 vector double vec_doublee (vector unsigned int);
29748 vector double vec_doublee (vector float);
29749 vector double vec_doubleh (vector signed int);
29750 vector double vec_doubleh (vector unsigned int);
29751 vector double vec_doubleh (vector float);
29752 vector double vec_doublel (vector signed int);
29753 vector double vec_doublel (vector unsigned int);
29754 vector double vec_doublel (vector float);
29755 vector double vec_doubleo (vector signed int);
29756 vector double vec_doubleo (vector unsigned int);
29757 vector double vec_doubleo (vector float);.
29758 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
29759 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
29761 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
29762 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
29763 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
29765 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
29766 vec_doublel, vec_doubleh.
29767 * doc/extend.texi: Update the built-in documentation file for the
29768 new built-in functions.
29770 2017-06-02 David Malcolm <dmalcolm@redhat.com>
29773 * ipa-inline-analysis.c (free_growth_caches): Set
29774 edge_removal_hook_holder to NULL after removing it.
29776 2017-06-02 Sudakshina Das <sudi.das@arm.com>
29778 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
29779 comparision with zero.
29781 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29782 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29783 for early expansion of vec_min and vec_max builtins.
29784 (builtin_function_type): Add min/max unsigned variants to those
29785 identified as having unsigned arguments.
29787 2017-06-02 Olivier Hainque <hainque@adacore.com>
29789 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
29791 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29793 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
29794 Use VALL_F16 iterator rather than VALL.
29796 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29798 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
29799 Emit CBNZ inside loop when doing a strong exchange and comparing
29800 against zero. Generate the CC flags after the loop.
29802 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
29804 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
29805 (dl_section_ref): New.
29806 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
29807 On AIX, append an expression to subtract the size of the
29808 section length to dl_section_ref.
29810 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
29812 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
29813 for early expansion of vector absolute builtins.
29815 2017-06-02 Richard Biener <rguenther@suse.de>
29817 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
29818 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
29820 2017-06-02 Richard Biener <rguenther@suse.de>
29822 PR tree-optimization/80948
29823 * tree-tailcall.c (find_tail_calls): Track stmts to move in
29824 stmt order as well.
29826 2017-06-02 Richard Biener <rguenther@suse.de>
29828 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
29830 * tree-vect-stmts.c (process_use): Do not mark backedge defs
29831 for inductions as relevant.
29833 2017-06-02 Richard Biener <rguenther@suse.de>
29835 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
29836 (vectorizable_induction): ... this. Remove dead code.
29838 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
29840 * builtins. (expand_builtin_alloca): Remove second parameter and
29841 infer its value from the first parameter instead.
29842 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
29844 2017-06-02 Jakub Jelinek <jakub@redhat.com>
29846 PR rtl-optimization/80903
29847 * loop-doloop.c (add_test): Unshare sequence.
29849 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
29851 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
29853 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
29855 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
29857 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
29858 xlogue_layout::get_instance, logue_layout::xlogue_layout,
29859 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
29860 (xlogue_layout::get_stub_rtx): Make static.
29861 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
29862 (xlogue_layout::compute_stub_managed_regs): Rename to...
29863 (xlogue_layout::count_stub_managed_regs): ...this.
29864 (xlogue_layout::is_stub_managed_reg): New function.
29865 (xlogue_layout::m_stub_names): Rename to...
29866 (xlogue_layout::s_stub_names): ...this, make static.
29867 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
29868 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
29869 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
29870 xlogue_layout::s_stub_names): Instantiate statics.
29871 (stub_managed_regs): Remove.
29872 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
29873 (disable_call_ms2sysv_xlogues): Rename to...
29874 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
29875 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
29877 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
29878 change after reload_completed.
29879 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
29881 (ix86_expand_prologue): Likewise.
29882 (ix86_expand_epilogue): Likewise.
29883 (ix86_expand_split_stack_prologue): Likewise.
29884 (ix86_compute_frame_layout): Remove frame parameter ...
29885 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
29886 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
29888 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
29889 (ix86_frame): Move from here ...
29890 * config/i386/i386.h (ix86_frame): ... to here.
29891 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
29892 complete ix86_frame data structure instead. Remove some_ld_name.
29894 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29896 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
29897 symbols that hold a DECL_VALUE_EXPR.
29899 2017-06-01 Martin Jambor <mjambor@suse.cz>
29901 PR tree-optimization/80898
29902 * tree-sra.c (process_subtree_disqualification): Removed.
29903 (disqualify_candidate): Do not acll
29904 process_subtree_disqualification.
29905 (subtree_mark_written_and_enqueue): New function.
29906 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
29907 RHS has been disqualified and re-queue LHS if necessary. Apart
29908 from that, ignore disqualified RHS.
29910 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29912 * config/s390/s390.c (s390_emit_epilogue): Disable early return
29913 address fetch for z10 or later.
29915 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29917 * config/arc/arc.md (tst_movb): Add guard when splitting.
29919 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29921 * config/arc/arc.c (arc_can_eliminate): Test against
29922 arc_frame_pointer_needed.
29924 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29926 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
29927 to prevent store reordering.
29928 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
29929 (type): Add block type.
29930 (stack_tie): Define special instruction to be used in
29933 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29935 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
29936 constraint. It is not valid for the pattern.
29937 (noncommutative_binary_comparison): Likewise.
29939 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29941 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
29944 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29946 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
29947 be used by the reg-alloc.
29949 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29951 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
29952 reg-alloc when having mul64 or mul32x16 instructions.
29953 (mulsidi3): Likewise.
29954 (umulsidi3): Likewise.
29955 (mulsi32x16): New pattern.
29956 (mulsi64): Likewise.
29957 (mulsidi64): Likewise.
29958 (umulsidi64): Likewise.
29959 (MUL32x16_REG): Define.
29960 (mul64_600): Use MUL32x16_REG.
29961 (mac64_600): Likewise.
29962 (umul64_600): Likewise.
29963 (umac64_600): Likewise.
29965 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
29967 * config/arc/arc.md (mulsi3_700): Make it commutative.
29969 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
29971 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
29973 (*sign_extendsidi2_insn): Likewise for movstosw.
29975 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
29977 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
29978 the type of the input discriminant value. Convert the
29979 discriminant value of signedness vary.
29981 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
29983 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
29984 Add to -Wall section.
29986 2017-06-01 Richard Biener <rguenther@suse.de>
29988 PR middle-end/66313
29989 * fold-const.c (fold_plusminus_mult_expr): If the factored
29990 factor may be zero use a wrapping type for the inner operation.
29991 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
29992 and handle moved defs.
29993 (process_assignment): Properly guard the unary op case. Return a
29994 tri-state indicating that moving the stmt before the call may allow
29995 to continue. Pass through to_move.
29996 (find_tail_calls): Handle moving unrelated defs before
29999 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
30002 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
30003 splitter result in the canonical way.
30005 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
30007 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
30008 also for 32bit target. Update insn attributes.
30009 (zero-extendsidi2 splitter): Allow all registers for operand 1.
30011 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
30013 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
30014 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
30015 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
30016 (_mm_maskz_min_ss): New intrinsics.
30018 2017-05-31 Martin Liska <mliska@suse.cz>
30020 * tree-vect-loop.c (vect_create_epilog_for_reduction):
30021 Change comment style to one we normally use.
30022 (vectorizable_reduction): Likewise.
30023 (vectorizable_induction): Likewise.
30024 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
30025 (vectorizable_call): Likewise.
30026 (vectorizable_simd_clone_call): Likewise.
30027 (vectorizable_conversion): Likewise.
30028 (vectorizable_assignment): Likewise.
30029 (vectorizable_shift): Likewise.
30030 (vectorizable_operation): Likewise.
30031 (vectorizable_store): Likewise.
30032 (vectorizable_load): Likewise.
30033 * tree-vectorizer.h: Likewise.
30035 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
30037 * passes.c (emergency_dump_function): New.
30038 * tree-pass.h (emergency_dump_function): Declare.
30039 * plugin.c (plugins_internal_error_function): Remove.
30040 * plugin.h (plugins_internal_error_function): Remove declaration.
30041 * toplev.c (internal_error_function): New static function. Use it...
30042 (general_init): ...here.
30044 2017-05-31 Graham Markall <graham.markall@embecosm.com>
30046 * config/arc/arc.c (arc_print_operand): Handle constant operands.
30047 (arc_rtx_costs): Add costs for new patterns.
30048 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
30049 * config/arc/predicates.md: Add _1_2_3_operand predicate.
30051 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
30053 * tree-ssa-strlen.c (get_next_strinfo): New function.
30054 (get_stridx_plus_constant): Use it.
30055 (zero_length_string): Likewise.
30056 (adjust_related_strinfos): Likewise.
30057 (adjust_last_stmt): Likewise.
30059 2017-05-31 Richard Biener <rguenther@suse.de>
30062 * config/i386/i386.c (ix86_expand_builtin): Remove assert
30063 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
30065 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
30067 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
30068 loop_vinfo argument and use of dependence distance vectors.
30069 Check instead whether the two references differ only in their
30070 initial value and assume that they have the same alignment if the
30071 difference is a multiple of the vector alignment.
30072 (vect_analyze_data_refs_alignment): Update call accordingly.
30074 2017-05-31 Martin Liska <mliska@suse.cz>
30077 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
30079 2017-05-31 Bin Cheng <bin.cheng@arm.com>
30081 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
30082 (create_intersect_range_checks): Move from ...
30083 * tree-data-ref.c (create_intersect_range_checks_index)
30084 (create_intersect_range_checks): ... to here.
30085 (create_runtime_alias_checks): New function factored from ...
30086 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
30087 here. Call above function.
30088 * tree-data-ref.h (create_runtime_alias_checks): New function.
30090 2017-05-31 Bin Cheng <bin.cheng@arm.com>
30092 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
30093 segment length for dr_b and compute it in wide_int.
30095 2017-05-31 Richard Biener <rguenther@suse.de>
30097 PR tree-optimization/80906
30098 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
30099 and pass through iv_map.
30100 (copy_bb_and_scalar_dependences): Adjust.
30101 (translate_pending_phi_nodes): Likewise.
30102 (copy_loop_close_phi_args): Handle code-generating IVs instead
30105 2017-05-30 David Malcolm <dmalcolm@redhat.com>
30107 * diagnostic-color.c (color_dict): Add "type-diff".
30108 (parse_gcc_colors): Update comment.
30109 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
30110 -fdiagnostics-show-template-tree and -fno-elide-type.
30111 (GCC_COLORS): Add type-diff to example.
30113 (-fdiagnostics-show-template-tree): New.
30114 (-fno-elide-type): New.
30115 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
30116 the pp_format_decoder callback. Call any m_format_postprocessor's
30118 (pretty_printer::pretty_printer): Initialize
30119 m_format_postprocessor.
30120 (pretty_printer::~pretty_printer): Delete any
30121 m_format_postprocessor.
30122 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
30123 (class format_postprocessor): New class.
30124 (struct pretty_printer::format_decoder): Document the new parameters.
30125 (struct pretty_printer::m_format_postprocessor): New field.
30126 * tree-diagnostic.c (default_tree_printer): Update for new
30127 bool and const char ** params.
30128 * tree-diagnostic.h (default_tree_printer): Likewise.
30130 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
30132 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
30133 (lwa_operand): Delete rs6000_gen_cell_microcode test.
30134 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
30135 rs6000_gen_cell_microcode code.
30136 (rs6000_final_prescan_insn): Delete.
30137 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
30138 "warn-cell-microcode" entries.
30139 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
30140 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
30141 throughout. Change cc_reg_not_micro_cr0_operand to
30142 cc_reg_not_cr0_operand throughout.
30143 (*extendhi<mode>2_noload): Delete.
30144 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
30145 (mwarn-cell-microcode): Delete.
30146 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
30147 -mgen-cell-microcode and -mwarn-cell-microcode.
30149 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
30152 * config/i386/constraints.md (Yd): New constraint.
30154 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
30155 and (?Yd, r) alternatives. Update insn attributes.
30156 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
30157 and (?*Yd, r) alternatives. Update insn attributes.
30158 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
30160 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
30162 * gimplify.c (gimplify_modify_expr): Don't create a
30163 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
30166 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
30168 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
30170 2017-05-30 Richard Biener <rguenther@suse.de>
30172 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
30173 and reduc_def fields.
30174 (STMT_VINFO_REDUC_TYPE): New define.
30175 (STMT_VINFO_REDUC_DEF): Likewise.
30176 (vect_force_simple_reduction): Adjust prototype.
30177 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
30178 (vect_is_simple_reduction): Remove check_reduction argument.
30179 (vect_force_simple_reduction): Adjust and set
30180 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30181 (vectorizable_reduction): Do not re-do reduction analysis
30182 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
30183 * tree-parloops.c (gather_scalar_reductions): Adjust.
30185 2017-05-30 Richard Biener <rguenther@suse.de>
30187 PR middle-end/80901
30188 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
30191 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
30193 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
30194 Introduce unknown_misalignment parameter and remove vf.
30195 (vect_peeling_hash_get_lowest_cost):
30196 Pass unknown_misalignment parameter.
30197 (vect_enhance_data_refs_alignment):
30198 Fix unsupportable data ref treatment.
30200 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30202 * tree-vect-data-refs.c (vect_get_data_access_cost):
30203 Workaround for SLP handling.
30204 (vect_enhance_data_refs_alignment):
30205 Compute costs for doing no peeling at all, compare to the best
30206 peeling costs so far and avoid peeling if cheaper.
30208 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30210 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
30211 Return peeling info and set costs to zero for unlimited cost
30213 (vect_enhance_data_refs_alignment): Also inspect all datarefs
30214 with unknown misalignment. Compute and costs for unknown
30215 misalignment, compare them to the costs for known misalignment
30216 and choose the cheapest for peeling.
30218 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30220 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
30221 (vect_get_peeling_costs_all_drs): Create function.
30222 (vect_peeling_hash_get_lowest_cost):
30223 Use vect_get_peeling_costs_all_drs.
30224 (vect_peeling_supportable): Create function.
30225 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
30227 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
30229 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
30230 DR_HAS_NEGATIVE_STEP.
30231 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
30232 (vect_enhance_data_refs_alignment): Use.
30233 (vect_duplicate_ssa_name_ptr_info): Use.
30234 * tree-vectorizer.h (dr_misalignment): Use.
30235 (known_alignment_for_access_p): Use.
30237 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
30240 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
30242 (has_section_name): New function.
30244 2017-05-30 Martin Liska <mliska@suse.cz>
30247 * auto-profile.c (get_function_decl_from_block): Fix
30250 2017-05-30 Richard Biener <rguenther@suse.de>
30252 PR middle-end/80876
30253 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
30255 2017-05-30 Martin Liska <mliska@suse.cz>
30257 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
30258 * dumpfile.h (struct dump_file_info): Remove ctors.
30260 2017-05-30 Martin Liska <mliska@suse.cz>
30262 * predict.def: Fix GNU coding style.
30264 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30266 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
30267 Mark 'to' argument with ATTRIBUTE_UNUSED.
30269 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
30271 * config/xtensa/xtensa.c (xtensa_emit_call): Use
30272 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
30273 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
30276 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
30278 * doc/install.texi (Options specification): Restore entry of
30279 --enable-sjlj-exceptions.
30281 2017-05-27 Michael Eager <eager@eagercon.com>
30284 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
30286 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
30288 * config/microblaze/microblaze.h
30289 (FIXED_REGISTERS): Update in macro.
30290 (CALL_USED_REGISTERS): Update in macro.
30292 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
30294 * doc/install.texi: Add links to macOS binary distributions.
30296 2017-05-27 Jakub Jelinek <jakub@redhat.com>
30300 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30302 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30304 2017-05-26 Martin Liska <mliska@suse.cz>
30306 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
30308 2017-05-26 Martin Liska <mliska@suse.cz>
30310 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
30311 always leading ';; '.
30312 (dump_bb_info): Likewise.
30313 (brief_dump_cfg): Likewise.
30314 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
30315 * dumpfile.c: Remove usage of TDF_VERBOSE.
30316 * dumpfile.h (enum dump_kind): Likewise.
30317 (dump_gimple_bb_header): Do not use TDF_COMMENT.
30318 * print-tree.c (debug_verbose): Remove.
30319 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
30320 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
30321 * tree-diagnostic.c (default_tree_printer): Replace
30322 TDF_DIAGNOSTIC with TDF_SLIM.
30324 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30326 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
30327 in parameter loop, rather than loop_vinfo.
30328 (create_intersect_range_checks): Ditto.
30329 (vect_create_cond_for_alias_checks): Update call to above functions.
30331 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30333 PR tree-optimization/80815
30334 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
30335 for merging runtime alias checks. Handle negative DR_STEPs.
30337 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30339 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
30341 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
30342 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
30343 out code pruning runtime alias checks.
30344 * tree-data-ref.c (prune_runtime_alias_test_list): New function
30345 factored out from above.
30346 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30348 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
30350 (prune_runtime_alias_test_list): New decalaration.
30352 2017-05-26 Bin Cheng <bin.cheng@arm.com>
30354 * tree-vect-data-refs.c (compare_tree): Rename and move ...
30355 * tree-data-ref.c (data_ref_compare_tree): ... to here.
30356 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
30357 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
30358 (operator==, comp_dr_with_seg_len_pair): Ditto.
30359 (vect_prune_runtime_alias_test_list): Ditto.
30361 2017-05-26 Martin Liska <mliska@suse.cz>
30364 * params.def: Bound partial-inlining-entry-probability param.
30366 2017-05-26 Marek Polacek <polacek@redhat.com>
30369 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
30372 2017-05-26 Richard Biener <rguenther@suse.de>
30374 PR tree-optimization/80842
30375 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
30378 2017-05-26 Richard Biener <rguenther@suse.de>
30380 PR tree-optimization/80844
30381 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
30383 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
30385 * doc/md.texi (Machine Constraints): Update x86 family
30386 machine constraints section to match 'config/i386/constraints.md'.
30388 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
30390 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
30392 2017-05-25 Nathan Sidwell <nathan@acm.org>
30394 * doc/invoke.texi (--enable-languages): Update documentation.
30396 2017-05-25 Martin Liska <mliska@suse.cz>
30398 * dumpfile.c: Add TDF_FOLDING.
30399 * dumpfile.h (enum dump_kind): Likewise.
30400 * genmatch.c (dt_simplify::gen_1): Use it.
30402 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30404 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
30406 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30408 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
30409 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
30411 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
30413 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
30414 * match.pd (X == C): Rewrite it here.
30415 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
30416 with_certain_nonzero_bits2): New predicates.
30417 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
30419 2017-05-24 Nathan Sidwell <nathan@acm.org>
30421 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
30424 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
30427 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30429 * config/powerpcspe: New port. Files are copied from the rs6000
30430 port, with "rs6000" in filenames replaced by "powerpcspe".
30432 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
30434 PR rtl-optimization/80754
30435 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
30437 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
30439 * config/sparc/sparc.md (length): Return the correct value for -mflat
30440 sibcalls to match output_sibcall.
30442 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
30446 * config/rs6000/rs6000.c (struct machine_function): Add new field
30448 (rs6000_get_separate_components): Init that field, use it.
30449 (rs6000_components_for_bb): Use the field.
30451 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
30453 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
30455 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
30457 PR middle-end/80823
30458 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
30460 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30463 * config/s390/s390.c (s390_check_qrst_address): Check incoming
30464 address against address_operand predicate.
30465 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
30467 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
30469 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
30470 parameters passed indirectly.
30472 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
30474 * config/i386/i386.md (*movdi_internal): Remove SSE4
30475 alternative 18 (?r, *v). Update insn attributes.
30476 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
30477 Update insn attributes.
30478 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
30479 Update insn attributes.
30480 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
30481 alternative 1 (r, v). Remove isa attribute.
30482 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
30483 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
30484 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
30486 2017-05-23 Tom de Vries <tom@codesourcery.com>
30488 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
30491 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30493 * cgraphunit.c (symbol_table::process_new_functions): Update.
30494 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
30495 (inline_generate_summary): Rename to ...
30496 (ipa_fn_summary_generate): ... this one.
30497 (inline_read_summary): Rename to ...
30498 (ipa_fn_summary_read): ... this one.
30499 (inline_write_summary): Rename to ...
30500 (ipa_fn_summary_write): ... this one.
30501 (inline_free_summary): Rename to ...
30502 (ipa_free_fn_summary): ... this one.
30503 (pass_data_local_fn_summary, pass_local_fn_summary,
30504 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
30505 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
30506 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
30507 make_pass_ipa_fn_summary): New.
30508 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
30509 inline_write_summary, inline_free_summary): Remove.
30510 (ipa_free_fn_summary) : New.
30511 * ipa-inline.c (ipa_inline): Update.
30512 (pass_ipa_inline): Do not generate summaries.
30513 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
30515 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
30516 and add pass_ipa_fn_summary.
30517 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
30519 (make_pass_inline_parameters): Remove.
30521 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
30523 * omp-low.c (struct omp_context): Remove "default_kind" member.
30526 * omp-offload.c (execute_oacc_device_lower): Remove the
30527 parallelism dimensions function attributes for unparallelized
30528 OpenACC kernels constructs.
30530 2017-05-23 Martin Liska <mliska@suse.cz>
30532 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
30534 (cgraph_edge::make_speculative): Likewise.
30535 (cgraph_edge::resolve_speculation): Likewise.
30536 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30537 (cgraph_node::dump): Likewise.
30538 * cgraph.h: Likewise.
30539 * cgraphunit.c (analyze_functions): Likewise.
30540 (symbol_table::compile): Likewise.
30541 * ipa-cp.c (print_all_lattices): Likewise.
30542 (determine_versionability): Likewise.
30543 (initialize_node_lattices): Likewise.
30544 (ipcp_verify_propagated_values): Likewise.
30545 (estimate_local_effects): Likewise.
30546 (update_profiling_info): Likewise.
30547 (create_specialized_node): Likewise.
30548 (perhaps_add_new_callers): Likewise.
30549 (decide_about_value): Likewise.
30550 (decide_whether_version_node): Likewise.
30551 (identify_dead_nodes): Likewise.
30552 (ipcp_store_bits_results): Likewise.
30553 * ipa-devirt.c (dump_targets): Likewise.
30554 (ipa_devirt): Likewise.
30555 * ipa-icf.c (sem_item::dump): Likewise.
30556 (sem_function::equals): Likewise.
30557 (sem_variable::equals): Likewise.
30558 (sem_item_optimizer::read_section): Likewise.
30559 (sem_item_optimizer::execute): Likewise.
30560 (congruence_class::dump): Likewise.
30561 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
30562 (dump_inline_summary): Likewise.
30563 (estimate_node_size_and_time): Likewise.
30564 (inline_analyze_function): Likewise.
30565 * ipa-inline-transform.c (inline_call): Likewise.
30566 * ipa-inline.c (report_inline_failed_reason): Likewise.
30567 (want_early_inline_function_p): Likewise.
30568 (edge_badness): Likewise.
30569 (update_edge_key): Likewise.
30570 (inline_small_functions): Likewise.
30571 * ipa-profile.c (ipa_profile): Likewise.
30572 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
30573 (ipa_make_edge_direct_to_target): Likewise.
30574 (remove_described_reference): Likewise.
30575 (ipa_impossible_devirt_target): Likewise.
30576 (propagate_controlled_uses): Likewise.
30577 (ipa_print_node_params): Likewise.
30578 (ipcp_transform_function): Likewise.
30579 * ipa-pure-const.c (pure_const_read_summary): Likewise.
30580 (propagate_pure_const): Likewise.
30581 * ipa-reference.c (generate_summary): Likewise.
30582 (read_write_all_from_decl): Likewise.
30583 (propagate): Likewise.
30584 (ipa_reference_read_optimization_summary): Likewise.
30585 * ipa-utils.c (ipa_merge_profiles): Likewise.
30586 * ipa.c (walk_polymorphic_call_targets): Likewise.
30587 (symbol_table::remove_unreachable_nodes): Likewise.
30588 (ipa_single_use): Likewise.
30589 * passes.c (execute_todo): Likewise.
30590 * predict.c (drop_profile): Likewise.
30591 * symtab.c (symtab_node::get_dump_name): New function.
30592 (symtab_node::dump_name): Likewise.
30593 (symtab_node::dump_asm_name): Likewise.
30594 (symtab_node::dump_references): Likewise.
30595 (symtab_node::dump_referring): Likewise.
30596 (symtab_node::dump_base): Likewise.
30597 (symtab_node::debug_symtab): Likewise.
30598 * tree-sra.c (convert_callers_for_node): Likewise.
30599 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
30600 * value-prof.c (init_node_map): Likewise.
30602 2017-05-23 Martin Liska <mliska@suse.cz>
30604 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
30605 and symtab_node::debug_symtab to symbol_table::debug.
30606 * cgraphunit.c (analyze_functions): Use the renamed function.
30607 (symbol_table::compile): Likewise.
30608 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
30609 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
30610 * passes.c (execute_todo): Likewise.
30611 * symtab.c (symbol_table::dump): New function.
30612 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
30614 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30616 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
30617 that nonconst implies exec.
30619 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
30621 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30622 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30623 (inline_edge_summary_vec): Turn into ...
30624 (ipa_call_summaries): ... this one.
30625 (redirect_to_unreachable, edge_set_predicate,
30626 evaluate_properties_for_edge, inline_summary_alloc,
30627 reset_ipa_call_summary, reset_inline_summary,
30628 inline_summary_t::duplicate): Update.
30629 (inline_edge_duplication_hook): Turn to ...
30630 (ipa_call_summary_t::duplicate): ... this one.
30631 (inline_edge_removal_hook): Turn to ...
30632 (ipa_call_summary_t::remove): ... this one.
30633 (dump_inline_edge_summary): Turn to ...
30634 (dump_ipa_call_summary): ... this one.
30635 (estimate_function_body_sizes): Update.
30636 (inline_update_callee_summaries): Update.
30637 (remap_edge_change_prob): Update.
30638 (remap_edge_summaries): Update.
30639 (inline_merge_summary): Update.
30640 (do_estimate_edge_time): Update.
30641 (inline_generate_summary): Update.
30642 (inline_read_section): Update.
30643 (inline_read_summary): Update.
30644 (inline_free_summary): Update.
30645 * ipa-inline.c (can_inline_edge_p): Update.
30646 (compute_inlined_call_time): Update.
30647 (want_inline_small_function_p): Update.
30648 (edge_badness): Update.
30649 (early_inliner): Update.
30650 * ipa-inline.h (inline_edge_summary): Turn to ...
30651 (ipa_call_summary): ... this one.
30652 (ipa_call_summary_t): New class.
30653 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30654 (ipa_call_summaries): New.
30655 (inline_edge_summary): Remove.
30656 (estimate_edge_growth): Update.
30657 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30658 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30659 * ipa-split.c (execute_split_functions): Update.
30660 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30662 2017-05-23 Tom de Vries <tom@codesourcery.com>
30664 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30665 attributes): Document rdrand effective target.
30667 2017-05-23 Tom de Vries <tom@codesourcery.com>
30669 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
30670 attributes): Sort alphabetically.
30672 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
30674 * config/avr/genmultilib.awk: Use gsub instead of gensub.
30676 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
30679 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
30680 V2DF/V2DI splat into two separate patterns, one that handles
30681 registers, and the other that only handles memory. Drop support
30682 for splatting from a GPR on ISA 2.07 and then splitting the
30683 splat into direct move and splat.
30684 (vsx_splat_<mode>_reg): Likewise.
30685 (vsx_splat_<mode>_mem): Likewise.
30687 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
30689 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
30691 2017-05-22 Jakub Jelinek <jakub@redhat.com>
30693 PR middle-end/80809
30694 * omp-low.c (finish_taskreg_remap): New function.
30695 (finish_taskreg_scan): If unit size of ctx->record_type
30696 is non-constant, unshare the size expression and replace
30697 decls in it with possible outer var refs.
30699 PR middle-end/80809
30700 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
30701 GOVD_SHARED rather than GOVD_PRIVATE with it.
30702 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
30703 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
30705 PR middle-end/80853
30706 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
30707 as last argument to build_outer_var_ref for pointer bases of array
30708 section reductions.
30710 2017-05-19 Martin Sebor <msebor@redhat.com>
30712 * print-tree.c (print_node): Print DECL_READ_P flag.
30714 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30716 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
30717 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
30718 * cgraph.c: Likewise.
30719 * cgraphunit.c: Likewise.
30720 * gengtype.c: Likewise.
30721 * ipa-cp.c: Likewise.
30722 * ipa-devirt.c: Likewise.
30723 * ipa-icf.c: Likewise.
30724 * ipa-predicate.c: Likewise.
30725 * ipa-profile.c: Likewise.
30726 * ipa-prop.c: Likewise.
30727 * ipa-split.c: Likewise.
30729 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
30730 edge_predicate_pool, dump_inline_hints,
30731 inline_summary::account_size_time, redirect_to_unreachable,
30732 edge_set_predicate, set_hint_predicate,
30733 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
30734 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
30735 inline_summary_t::remove, remap_hint_predicate_after_duplication,
30736 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
30737 ipa_call_summary_t::remove, initialize_growth_caches,
30738 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
30739 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
30740 mark_modified, unmodified_parm_1, unmodified_parm,
30741 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
30742 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
30743 compute_bb_predicates, will_be_nonconstant_expr_predicate,
30744 will_be_nonconstant_predicate, record_modified_bb_info,
30745 get_minimal_bb, record_modified, param_change_prob,
30746 phi_result_unknown_predicate, predicate_for_phi_result,
30747 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
30748 estimate_function_body_sizes, compute_inline_parameters,
30749 compute_inline_parameters_for_curren, pass_data_inline_parameters,
30750 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
30751 inline_update_callee_summaries, remap_edge_change_prob,
30752 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
30753 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
30754 inline_analyze_function, inline_summary_t::insert,
30755 inline_generate_summary, read_ipa_call_summary, inline_read_section,
30756 inline_read_summary, write_ipa_call_summary, inline_write_summary,
30757 inline_free_summary): Move to ipa-fnsummary.h
30758 (predicate_t): Remove.
30759 * ipa-fnsummary.c: New file.
30760 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
30761 (enum inline_hints_vals, inline_hints, agg_position_info,
30762 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
30763 inline_summaries, ipa_call_summary, ipa_call_summary_t,
30764 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
30765 dump_inline_summary, dump_inline_hints, inline_generate_summary,
30766 inline_read_summary, inline_write_summary, inline_free_summary,
30767 inline_analyze_function, initialize_inline_failed,
30768 inline_merge_summary, inline_update_overall_summary,
30769 compute_inline_parameters): Move to ipa-fnsummary.h
30770 * ipa-fnsummary.h: New file.
30771 * ipa-inline-transform.h: Include ipa-inline.h.
30772 * ipa-inline.c: LIkewise.
30774 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30776 * ipa-inline.c (edge_badness): Use inlined_time instead of
30777 inline_summaries->get.
30779 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30781 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
30783 2017-05-22 Nathan Sidwell <nathan@acm.org>
30785 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
30786 (fdump-lang): Document 'raw' option.
30787 * dumpfile.h (TDI_tu): Delete.
30788 * dumpfile.c (dump_files): Remove translation-unit.
30789 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
30791 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
30793 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
30794 command option from $(AWK) call.
30795 * config/avr/genmultilib.awk: Simplify and rewrite so that it
30796 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
30797 [FORMAT]: Remove handling of variable.
30798 * config/avr/t-multilib: Regenerate.
30800 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30802 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
30804 (dump_inline_summary): Do not print self_time.
30805 (estimate_function_body_sizes): Do not set self_time.
30806 (compute_inline_parameters): Likewise.
30807 (inline_read_section, inline_write_summary): Do not stream self_time.
30808 * ipa-inline.h (inline_summary): Drop self_time.
30810 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
30812 * ipa-inline-analysis.c (account_size_time): Rename to ...
30813 (inline_summary::account_size_time): ... this one.
30814 (reset_ipa_call_summary): Turn to ...
30815 (ipa_call_summary::reset): ... this one.
30816 (reset_inline_summary): Turn to ...
30817 (inline_summary::reset): ... this one.
30818 (inline_summary_t::remove): Update.
30819 (inline_summary_t::duplicate): Update.
30820 (ipa_call_summary_t::remove): Update.
30821 (dump_inline_summary): Update.
30822 (estimate_function_body_sizes): Update.
30823 (compute_inline_parameters): Update.
30824 (estimate_node_size_and_time): Update.
30825 (inline_merge_summary): Update.
30826 (inline_update_overall_summary): Update.
30827 (inline_read_section): Update.
30828 (inline_write_summary): Update.
30829 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
30830 add account_size_time and reset member functions.
30831 (ipa_call_summary): Add reset function.
30832 * ipa-predicate.h (predicate::operator &): Constify.
30834 2017-05-22 Richard Biener <rguenther@suse.de>
30836 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
30838 2017-05-19 Jason Merrill <jason@redhat.com>
30840 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
30842 2017-05-19 Marek Polacek <polacek@redhat.com>
30845 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
30846 TYPE_OVERFLOW_WRAPS checks.
30848 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
30850 * tree-core.h (enum omp_clause_default_kind): Add
30851 "OMP_CLAUSE_DEFAULT_PRESENT".
30852 * tree-pretty-print.c (dump_omp_clause): Handle it.
30853 * gimplify.c (enum gimplify_omp_var_data): Add
30854 "GOVD_MAP_FORCE_PRESENT".
30855 (gimplify_adjust_omp_clauses_1): Map it to
30856 "GOMP_MAP_FORCE_PRESENT".
30857 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
30859 * gimplify.c (oacc_default_clause): Clarify.
30861 2017-05-19 Nathan Sidwell <nathan@acm.org>
30863 LANG_HOOK_REGISTER_DUMPS
30864 * toplev.c (general_init): Call register dump lang hook.
30865 * doc/invoke.texi: Document -fdump-lang option family.
30866 * dumpfile.c (dump_files): Remove class dump here.
30867 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
30868 * dumpfile.h (tree_dump_index): Remove TDI_class.
30869 * langhooks-def.h (lhd_register_dumps): Declare.
30870 (LANG_HOOKS_REGISTER_DUMPS): Define.
30871 (LANG_HOOKS_INITIALIZER): Add it.
30872 * langhooks.c (lhd_register_dumps): Define.
30873 * langhooks.h (struct lang_hooks): Add register_dumps.
30875 2017-05-19 Nathan Sidwell <nathan@acm.org>
30877 * context.h (context::set_passes): New.
30878 * context.c (context::context): Do not create pass manager.
30879 * toplev.c (general_init): Create pass manager here.
30881 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
30883 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
30884 use this splitter if two add or or instructions would also work for
30885 the constant we want to generate.
30887 2017-05-19 Richard Biener <rguenther@suse.de>
30890 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
30891 predicate evaluation.
30893 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30895 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
30897 * ipa-inline.c (want_inline_small_function_p): Do not cast to
30900 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
30902 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
30903 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
30904 (inline_edge_summary_vec): Turn into ...
30905 (ipa_call_summaries): ... this one.
30906 (redirect_to_unreachable, edge_set_predicate,
30907 evaluate_properties_for_edge, inline_summary_alloc,
30908 reset_ipa_call_summary, reset_inline_summary,
30909 inline_summary_t::duplicate): Update.
30910 (inline_edge_duplication_hook): Turn to ...
30911 (ipa_call_summary_t::duplicate): ... this one.
30912 (inline_edge_removal_hook): Turn to ...
30913 (ipa_call_summary_t::remove): ... this one.
30914 (dump_inline_edge_summary): Turn to ...
30915 (dump_ipa_call_summary): ... this one.
30916 (estimate_function_body_sizes): Update.
30917 (inline_update_callee_summaries): Update.
30918 (remap_edge_change_prob): Update.
30919 (remap_edge_summaries): Update.
30920 (inline_merge_summary): Update.
30921 (do_estimate_edge_time): Update.
30922 (inline_generate_summary): Update.
30923 (inline_read_section): Update.
30924 (inline_read_summary): Update.
30925 (inline_free_summary): Update.
30926 * ipa-inline.c (can_inline_edge_p): Update.
30927 (compute_inlined_call_time): Update.
30928 (want_inline_small_function_p): Update.
30929 (edge_badness): Update.
30930 (early_inliner): Update.
30931 * ipa-inline.h (inline_edge_summary): Turn to ...
30932 (ipa_call_summary): ... this one.
30933 (ipa_call_summary_t): New class.
30934 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
30935 (ipa_call_summaries): New.
30936 (inline_edge_summary): Remove.
30937 (estimate_edge_growth): Update.
30938 * ipa-profile.c (ipa_propagate_frequency_1): Update.
30939 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
30940 * ipa-split.c (execute_split_functions): Update.
30941 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
30943 2017-05-19 Richard Biener <rguenther@suse.de>
30945 PR middle-end/80764
30946 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
30948 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
30950 * config/rs6000/rs6000.c (struct machine_function): Add field
30951 fpr_is_wrapped_separately.
30952 (rs6000_get_separate_components): Use 64 components. Handle the
30953 new FPR components.
30954 (rs6000_components_for_bb): Handle the FPR components.
30955 (rs6000_emit_prologue_components): Handle the FPR components.
30956 (rs6000_emit_epilogue_components): Handle the FPR components.
30957 (rs6000_set_handled_components): Handle the FPR components.
30958 (rs6000_emit_prologue): Don't output prologue code for those FPRs
30959 that are already separately shrink-wrapped.
30960 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
30961 that are already separately shrink-wrapped.
30963 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
30966 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
30969 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
30970 (define_peephole2 for Altivec d-form load): Add peepholes to catch
30971 cases where the register allocator uses a move and an offsettable
30972 memory operation to/from a FPR register on ISA 2.06/2.07.
30973 (define_peephole2 for Altivec d-form store): Likewise.
30975 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
30978 * config/i386/mmx.md (*mov<mode>_internal): Enable
30979 alternatives 11, 12, 13 and 14 also for 32bit targets.
30980 Remove alternatives 15, 16, 17 and 18.
30981 * config/i386/sse.md (vec_concatv2di): Change
30982 alternative (!x, *y) to (x, ?!*Yn).
30984 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
30986 * dumpfile.h (enum dump_kind): Remove stray comma.
30988 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
30990 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
30991 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
30992 predicate::num_conditions
30993 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
30994 (CHANGED): turn into predicate::changed.
30995 (agg_position_info): Move to ipa-predicate.h
30996 (add_condition, predicate::add_clause, predicate::operator &=,
30997 predicate::or_with, predicate::evaluate, predicate::probability,
30998 dump_condition, dump_clause, predicate::dump,
30999 predicate::remap_after_duplication, predicate::remap_after_inlining,
31000 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
31001 (evaluate_conditions_for_known_args): Update.
31002 (set_cond_stmt_execution_predicate): Update.
31003 * ipa-inline.h: Include ipa-predicate.h
31004 (condition, inline_param_summary, conditions, agg_position_info,
31005 predicate): Move to ipa-predicate.h
31006 * ipa-predicate.c: New file.
31007 * ipa-predicate.h: New file.
31009 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
31011 * final.c (leaf_function_p): Check we are not in a sequence.
31013 2017-05-18 Martin Liska <mliska@suse.cz>
31015 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
31016 * dumpfile.c (dump_register): Use new enum dump_kind.
31017 (get_dump_file_name): Likewise.
31018 (dump_enable_all): Likewise.
31019 (dump_switch_p_1): Likewise.
31020 (enable_rtl_dump_file): Remove usage of TDF_RTL.
31021 * dumpfile.h (enum dump_kind): New enum type.
31022 (struct dump_file_info): Create constructor and
31023 format fields and comments.
31024 * passes.c (pass_manager::register_one_dump_file):
31026 * statistics.c (statistics_early_init): Likewise.
31027 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
31028 TDF_TREE with TDF_SLIM.
31029 (gather_memory_references_ref): Likewise.
31031 2017-05-18 Martin Liska <mliska@suse.cz>
31033 * vec.h (struct vnull): Use it.
31035 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
31037 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
31038 (true_predicate, false_predicate, true_predicate_p,
31039 false_predicate_p): Remove.
31040 (single_cond_predicate, not_inlined_predicate): Turn to member function
31042 (add_condition): Update.
31043 (add_clause): Turn to...
31044 (predicate::add_clause): ... this one; update; allow passing NULL
31046 (and_predicates): Turn to ...
31047 (predicate::operator &=): ... this one.
31048 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
31049 (or_predicates): Turn to ...
31050 (predicate::or_with): ... this one.
31051 (evaluate_predicate): Turn to ...
31052 (predicate::evaluate): ... this one.
31053 (predicate_probability): Turn to ...
31054 (predicate::probability): ... this one.
31055 (dump_condition): Update.
31056 (dump_predicate): Turn to ...
31057 (predicate::dump): ... this one.
31058 (account_size_time): Update.
31059 (edge_set_predicate): Update.
31060 (set_hint_predicate): UPdate.
31061 (evaluate_conditions_for_known_args): Update.
31062 (evaluate_properties_for_edge): Update.
31063 (remap_predicate_after_duplication): Turn to...
31064 (predicate::remap_after_duplication): ... this one.
31065 (remap_hint_predicate_after_duplication): Update.
31066 (inline_summary_t::duplicate): UPdate.
31067 (dump_inline_edge_summary): Update.
31068 (dump_inline_summary): Update.
31069 (set_cond_stmt_execution_predicate): Update.
31070 (set_switch_stmt_execution_predicate): Update.
31071 (compute_bb_predicates): Update.
31072 (will_be_nonconstant_expr_predicate): Update.
31073 (will_be_nonconstant_predicate): Update.
31074 (phi_result_unknown_predicate): Update.
31075 (predicate_for_phi_result): Update.
31076 (array_index_predicate): Update.
31077 (estimate_function_body_sizes): Update.
31078 (estimate_node_size_and_time): Update.
31079 (estimate_ipcp_clone_size_and_time): Update.
31080 (remap_predicate): Rename to ...
31081 (predicate::remap_after_inlining): ... this one.
31082 (remap_hint_predicate): Update.
31083 (inline_merge_summary): Update.
31084 (inline_update_overall_summary): Update.
31085 (estimate_size_after_inlining): Update.
31086 (read_predicate): Rename to ...
31087 (predicate::stream_in): ... this one.
31088 (read_inline_edge_summary): Update.
31089 (write_predicate): Rename to ...
31090 (predicate::stream_out): ... this one.
31091 (write_inline_edge_summary): Update.
31092 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
31093 (clause_t): Turn to uint32_t
31094 (predicate): Turn to class; implement constructor and operators
31096 (size_time_entry): Update.
31097 (inline_summary): Update.
31098 (inline_edge_summary): Update.
31100 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
31102 * fold-const.c (fold_binary_loc): Move transformation...
31103 * match.pd (C - X CMP X): ... here.
31105 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
31107 * config/sparc/sparc.c (sparc_option_override): Set function
31108 alignment for -mcpu=niagara7 to 64 to match the I$ line.
31109 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
31111 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
31113 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
31115 2017-05-18 Marek Polacek <polacek@redhat.com>
31118 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
31119 (pass_ubsan::execute): Call gimple_assign_single_p instead of
31120 gimple_assign_load_p.
31122 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
31124 PR middle-end/80692
31125 * real.c (do_compare): Give decimal_do_compare preference over
31126 comparing just the signs.
31128 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
31130 * doc/md.texi (Canonicalization of Instructions): Describe the
31131 canonical form of instructions that inherently set a condition
31134 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
31136 PR middle-end/80775
31137 * tree-cfg.c: Move deletion of unreachable case statements to after
31138 the merging of consecutive case labels.
31140 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
31142 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
31143 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
31144 restoring of callee-saved registers.
31146 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
31148 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
31149 * config/visium/visium.c (single_set_and_flags): Likewise.
31150 * config/visium/visium.md (Substitutions): Likewise.
31152 2017-05-17 Martin Liska <mliska@suse.cz>
31154 * cfg.c: Introduce dump_flags_t type and
31155 use it instead of int type.
31157 * cfghooks.c: Likewise.
31158 * cfghooks.h (struct cfg_hooks): Likewise.
31159 * cfgrtl.c: Likewise.
31160 * cfgrtl.h: Likewise.
31161 * cgraph.c (cgraph_node::get_body): Likewise.
31162 * coretypes.h: Likewise.
31163 * domwalk.c: Likewise.
31164 * domwalk.h: Likewise.
31165 * dumpfile.c (struct dump_option_value_info): Likewise.
31166 (dump_enable_all): Likewise.
31167 (dump_switch_p_1): Likewise.
31168 (opt_info_switch_p): Likewise.
31169 * dumpfile.h (enum tree_dump_index): Likewise.
31170 (struct dump_file_info): Likewise.
31171 * genemit.c: Likewise.
31172 * generic-match-head.c: Likewise.
31173 * gengtype.c (open_base_files): Likewise.
31174 * gimple-pretty-print.c: Likewise.
31175 * gimple-pretty-print.h: Likewise.
31176 * graph.c (print_graph_cfg): Likewise.
31177 * graphite-scop-detection.c (dot_all_sese): Likewise.
31178 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
31179 * loop-unroll.c (report_unroll): Likewise.
31180 * passes.c (pass_manager::register_one_dump_file): Likewise.
31181 * print-tree.c: Likewise.
31182 * statistics.c: Likewise.
31183 * tree-cfg.c: Likewise.
31184 * tree-cfg.h: Likewise.
31185 * tree-dfa.c: Likewise.
31186 * tree-dfa.h: Likewise.
31187 * tree-dump.c (dump_function): Likewise.
31188 * tree-dump.h (struct dump_info): Likewise.
31189 * tree-pretty-print.c: Likewise.
31190 * tree-pretty-print.h: Likewise.
31191 * tree-ssa-live.c: Likewise.
31192 * tree-ssa-live.h: Likewise.
31193 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
31194 * tree-vect-loop.c: Likewise.
31195 * tree-vect-slp.c: Likewise.
31197 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
31198 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31200 PR tree-optimization/80457
31201 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
31202 of all arguments to a statement as scalar_to_vec operations.
31203 (vectorizable_call): Adjust call to vect_model_simple_cost for
31205 (vectorizable_conversion): Likewise.
31206 (vectorizable_assignment): Likewise.
31207 (vectorizable_shift): Likewise.
31208 (vectorizable_operation): Likewise.
31209 (vectorizable_comparison): Likewise.
31210 (vect_is_simple_cond): Record the def types for operands.
31211 (vectorizable_condition): Likewise, call vect_model_simple_cost.
31212 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
31213 for statement argument count.
31215 2017-05-16 Carl Love <cel@us.ibm.com>
31217 * config/rs6000/rs6000-c: Add support for built-in functions
31218 vector unsigned long long vec_bperm (vector unsigned long long,
31219 vector unsigned char)
31220 vector signed long long vec_mule (vector signed int,
31222 vector unsigned long long vec_mule (vector unsigned int,
31223 vector unsigned int)
31224 vector signed long long vec_mulo (vector signed int,
31226 vector unsigned long long vec_mulo (vector unsigned int,
31227 vector unsigned int)
31228 vector signed char vec_sldw (vector signed char,
31229 vector signed char,
31231 vector unsigned char vec_sldw (vector unsigned char,
31232 vector unsigned char,
31234 vector signed short vec_sldw (vector signed short,
31235 vector signed short,
31237 vector unsigned short vec_sldw (vector unsigned short,
31238 vector unsigned short,
31240 vector signed int vec_sldw (vector signed int,
31243 vector unsigned int vec_sldw (vector unsigned int,
31244 vector unsigned int,
31246 vector signed long long vec_sldw (vector signed long long,
31247 vector signed long long,
31249 vector unsigned long long vec_sldw (vector unsigned long long,
31250 vector unsigned long long,
31252 * config/rs6000/rs6000-c: Add support for built-in functions
31253 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
31254 * config/rs6000/altivec.h: Add defintion for vec_sldw.
31255 * doc/extend.texi: Update the built-in documentation for the
31256 new built-in functions.
31258 2017-05-16 Marek Polacek <polacek@redhat.com>
31262 * tree.c (save_expr): Don't fold the expression.
31264 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
31266 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
31267 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
31268 and (?*y,m). Update insn attributes.
31270 2017-05-16 Martin Liska <mliska@suse.cz>
31272 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
31273 flags argument of print_gimple_stmt, print_gimple_expr,
31274 print_generic_stmt and print_generic_expr.
31275 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
31276 * coretypes.h: Likewise.
31277 * except.c (dump_eh_tree): Likewise.
31278 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
31279 * gimple-pretty-print.h: Likewise.
31280 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
31281 (backprop::push_to_worklist): Likewise.
31282 (backprop::pop_from_worklist): Likewise.
31283 (backprop::process_use): Likewise.
31284 (backprop::intersect_uses): Likewise.
31285 (note_replacement): Likewise.
31286 * gimple-ssa-store-merging.c
31287 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
31288 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
31289 (pass_store_merging::execute): Likewise.
31290 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
31291 (ssa_base_cand_dump_callback): Likewise.
31292 (dump_incr_vec): Likewise.
31293 (replace_refs): Likewise.
31294 (replace_mult_candidate): Likewise.
31295 (create_add_on_incoming_edge): Likewise.
31296 (create_phi_basis): Likewise.
31297 (insert_initializers): Likewise.
31298 (all_phi_incrs_profitable): Likewise.
31299 (introduce_cast_before_cand): Likewise.
31300 (replace_one_candidate): Likewise.
31301 * gimplify.c (gimplify_expr): Likewise.
31302 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
31303 (set_rename): Likewise.
31304 (rename_uses): Likewise.
31305 (copy_loop_phi_nodes): Likewise.
31306 (add_close_phis_to_merge_points): Likewise.
31307 (copy_loop_close_phi_args): Likewise.
31308 (copy_cond_phi_args): Likewise.
31309 (graphite_copy_stmts_from_block): Likewise.
31310 (translate_pending_phi_nodes): Likewise.
31311 * graphite-poly.c (print_pdr): Likewise.
31312 (dump_gbb_cases): Likewise.
31313 (dump_gbb_conditions): Likewise.
31314 (print_scop_params): Likewise.
31315 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
31316 (build_cross_bb_scalars_use): Likewise.
31317 (gather_bbs::before_dom_children): Likewise.
31318 * hsa-dump.c (dump_hsa_immed): Likewise.
31319 * ipa-cp.c (print_ipcp_constant_value): Likewise.
31320 (get_replacement_map): Likewise.
31321 * ipa-inline-analysis.c (dump_condition): Likewise.
31322 (estimate_function_body_sizes): Likewise.
31323 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
31324 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
31325 * ipa-prop.c (ipa_dump_param): Likewise.
31326 (ipa_print_node_jump_functions_for_edge): Likewise.
31327 (ipa_modify_call_arguments): Likewise.
31328 (ipa_modify_expr): Likewise.
31329 (ipa_dump_param_adjustments): Likewise.
31330 (ipa_dump_agg_replacement_values): Likewise.
31331 (ipcp_modif_dom_walker::before_dom_children): Likewise.
31332 * ipa-pure-const.c (check_stmt): Likewise.
31333 (pass_nothrow::execute): Likewise.
31334 * ipa-split.c (execute_split_functions): Likewise.
31335 * omp-offload.c (dump_oacc_loop_part): Likewise.
31336 (dump_oacc_loop): Likewise.
31337 * trans-mem.c (tm_log_emit): Likewise.
31338 (tm_memopt_accumulate_memops): Likewise.
31339 (dump_tm_memopt_set): Likewise.
31340 (dump_tm_memopt_transform): Likewise.
31341 * tree-cfg.c (gimple_verify_flow_info): Likewise.
31342 (print_loop): Likewise.
31343 * tree-chkp-opt.c (chkp_print_addr): Likewise.
31344 (chkp_gather_checks_info): Likewise.
31345 (chkp_get_check_result): Likewise.
31346 (chkp_remove_check_if_pass): Likewise.
31347 (chkp_use_outer_bounds_if_possible): Likewise.
31348 (chkp_reduce_bounds_lifetime): Likewise.
31349 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
31350 (chkp_mark_completed_bounds): Likewise.
31351 (chkp_register_incomplete_bounds): Likewise.
31352 (chkp_mark_invalid_bounds): Likewise.
31353 (chkp_maybe_copy_and_register_bounds): Likewise.
31354 (chkp_build_returned_bound): Likewise.
31355 (chkp_get_bound_for_parm): Likewise.
31356 (chkp_build_bndldx): Likewise.
31357 (chkp_get_bounds_by_definition): Likewise.
31358 (chkp_generate_extern_var_bounds): Likewise.
31359 (chkp_get_bounds_for_decl_addr): Likewise.
31360 * tree-chrec.c (chrec_apply): Likewise.
31361 * tree-data-ref.c (dump_data_reference): Likewise.
31362 (dump_subscript): Likewise.
31363 (dump_data_dependence_relation): Likewise.
31364 (analyze_overlapping_iterations): Likewise.
31365 * tree-inline.c (expand_call_inline): Likewise.
31366 (tree_function_versioning): Likewise.
31367 * tree-into-ssa.c (dump_defs_stack): Likewise.
31368 (dump_currdefs): Likewise.
31369 (dump_names_replaced_by): Likewise.
31370 (dump_update_ssa): Likewise.
31371 (update_ssa): Likewise.
31372 * tree-object-size.c (pass_object_sizes::execute): Likewise.
31373 * tree-parloops.c (build_new_reduction): Likewise.
31374 (try_create_reduction_list): Likewise.
31375 (ref_conflicts_with_region): Likewise.
31376 (oacc_entry_exit_ok_1): Likewise.
31377 (oacc_entry_exit_single_gang): Likewise.
31378 * tree-pretty-print.h: Likewise.
31379 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
31380 (get_scalar_evolution): Likewise.
31381 (add_to_evolution): Likewise.
31382 (get_loop_exit_condition): Likewise.
31383 (analyze_evolution_in_loop): Likewise.
31384 (analyze_initial_condition): Likewise.
31385 (analyze_scalar_evolution): Likewise.
31386 (instantiate_scev): Likewise.
31387 (number_of_latch_executions): Likewise.
31388 (gather_chrec_stats): Likewise.
31389 (final_value_replacement_loop): Likewise.
31390 (scev_const_prop): Likewise.
31391 * tree-sra.c (dump_access): Likewise.
31392 (disqualify_candidate): Likewise.
31393 (create_access): Likewise.
31394 (reject): Likewise.
31395 (maybe_add_sra_candidate): Likewise.
31396 (create_access_replacement): Likewise.
31397 (analyze_access_subtree): Likewise.
31398 (analyze_all_variable_accesses): Likewise.
31399 (sra_modify_assign): Likewise.
31400 (initialize_constant_pool_replacements): Likewise.
31401 (find_param_candidates): Likewise.
31402 (decide_one_param_reduction): Likewise.
31403 (replace_removed_params_ssa_names): Likewise.
31404 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
31405 * tree-ssa-copy.c (dump_copy_of): Likewise.
31406 (copy_prop_visit_cond_stmt): Likewise.
31407 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
31408 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
31409 (record_equivalences_from_stmt): Likewise.
31410 * tree-ssa-dse.c (compute_trims): Likewise.
31411 (delete_dead_call): Likewise.
31412 (delete_dead_assignment): Likewise.
31413 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
31414 (forward_propagate_into_cond): Likewise.
31415 (pass_forwprop::execute): Likewise.
31416 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
31417 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
31419 (move_computations_worker): Likewise.
31420 (execute_sm): Likewise.
31421 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
31422 (remove_exits_and_undefined_stmts): Likewise.
31423 (remove_redundant_iv_tests): Likewise.
31424 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
31425 (adjust_iv_update_pos): Likewise.
31426 * tree-ssa-math-opts.c (bswap_replace): Likewise.
31427 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
31428 (value_replacement): Likewise.
31429 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
31430 * tree-ssa-pre.c (print_pre_expr): Likewise.
31431 (get_representative_for): Likewise.
31432 (create_expression_by_pieces): Likewise.
31433 (insert_into_preds_of_block): Likewise.
31434 (eliminate_insert): Likewise.
31435 (eliminate_dom_walker::before_dom_children): Likewise.
31436 (eliminate): Likewise.
31437 (remove_dead_inserted_code): Likewise.
31438 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
31439 * tree-ssa-reassoc.c (get_rank): Likewise.
31440 (eliminate_duplicate_pair): Likewise.
31441 (eliminate_plus_minus_pair): Likewise.
31442 (eliminate_not_pairs): Likewise.
31443 (undistribute_ops_list): Likewise.
31444 (eliminate_redundant_comparison): Likewise.
31445 (update_range_test): Likewise.
31446 (optimize_range_tests_var_bound): Likewise.
31447 (optimize_vec_cond_expr): Likewise.
31448 (rewrite_expr_tree): Likewise.
31449 (rewrite_expr_tree_parallel): Likewise.
31450 (linearize_expr): Likewise.
31451 (break_up_subtract): Likewise.
31452 (linearize_expr_tree): Likewise.
31453 (attempt_builtin_powi): Likewise.
31454 (attempt_builtin_copysign): Likewise.
31455 (transform_stmt_to_copy): Likewise.
31456 (transform_stmt_to_multiply): Likewise.
31457 (dump_ops_vector): Likewise.
31458 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
31459 (print_scc): Likewise.
31460 (set_ssa_val_to): Likewise.
31461 (visit_reference_op_store): Likewise.
31462 (visit_use): Likewise.
31463 (sccvn_dom_walker::before_dom_children): Likewise.
31464 (run_scc_vn): Likewise.
31465 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
31467 (expr_hash_elt::print): Likewise.
31468 (const_and_copies::pop_to_marker): Likewise.
31469 (const_and_copies::record_const_or_copy_raw): Likewise.
31470 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
31471 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
31472 (dump_predicates): Likewise.
31473 (find_uninit_use): Likewise.
31474 (warn_uninitialized_phi): Likewise.
31475 (pass_late_warn_uninitialized::execute): Likewise.
31476 * tree-ssa.c (verify_vssa): Likewise.
31477 (verify_ssa): Likewise.
31478 (maybe_optimize_var): Likewise.
31479 * tree-vrp.c (dump_value_range): Likewise.
31480 (dump_all_value_ranges): Likewise.
31481 (dump_asserts_for): Likewise.
31482 (register_edge_assert_for_2): Likewise.
31483 (vrp_visit_cond_stmt): Likewise.
31484 (vrp_visit_switch_stmt): Likewise.
31485 (vrp_visit_stmt): Likewise.
31486 (vrp_visit_phi_node): Likewise.
31487 (simplify_cond_using_ranges_1): Likewise.
31488 (fold_predicate_in): Likewise.
31489 (evrp_dom_walker::before_dom_children): Likewise.
31490 (evrp_dom_walker::push_value_range): Likewise.
31491 (evrp_dom_walker::pop_value_range): Likewise.
31492 (execute_early_vrp): Likewise.
31494 2017-05-16 Richard Biener <rguenther@suse.de>
31496 * dwarf2out.c (loc_list_from_tree_1): Do not create
31497 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
31499 2017-05-16 Richard Biener <rguenther@suse.de>
31501 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
31503 (note_variable_value_in_expr): If we resolved the decl ref
31504 do not push to the stack.
31506 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
31508 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
31509 operations in fast-math mode.
31510 (vaddq_f16): Likewise.
31511 (vmul_f16): Likewise.
31512 (vmulq_f16): Likewise.
31513 (vsub_f16): Likewise.
31514 (vsubq_f16): Likewise.
31515 * config/arm/neon.md (add<mode>3): New.
31517 (fma:<VH:mode>3): New. Also remove outdated comment.
31520 2017-05-16 Martin Liska <mliska@suse.cz>
31524 * ipa-devirt.c (warn_types_mismatch): Fix typo.
31525 (odr_types_equivalent_p): Likewise.
31527 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
31529 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
31531 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
31534 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
31535 non-interunit SSE move alternatives with '?'.
31536 (zero-extendsidi peephole2): New peephole to skip intermediate
31537 general register in SSE zero-extend sequence.
31539 2017-05-15 Jeff Law <law@redhat.com>
31541 * reorg.c (relax_delay_slots): Create a new variable to hold
31542 the temporary target rather than clobbering TARGET_LABEL.
31544 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
31545 missing argument to extract_bit_field call.
31546 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
31548 2017-05-15 Martin Liska <mliska@suse.cz>
31551 * gcc.c (process_command): Do not allow empty argument of -o option.
31553 2017-05-15 Renlin Li <renlin.li@arm.com>
31555 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
31556 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
31557 * config/aarch64/constraints.md (Usf): Add long call check.
31558 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
31559 (call_value): Likewise.
31560 (sibcall): Likewise.
31561 (sibcall_value): Likewise.
31563 (call_value_insn): New.
31564 (sibcall_insn): Update rtx pattern.
31565 (sibcall_value_insn): Likewise.
31566 (call_internal): Remove.
31567 (call_value_internal): Likewise.
31568 (sibcall_internal): Likewise.
31569 (sibcall_value_internal): Likewise.
31570 (call_reg): Likewise.
31571 (call_symbol): Likewise.
31572 (call_value_reg): Likewise.
31573 (call_value_symbol): Likewise.
31575 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
31578 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
31580 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
31582 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
31583 compatible with CCGOCmode and with CCZmode.
31585 2017-05-14 Martin Sebor <msebor@redhat.com>
31587 PR middle-end/77671
31588 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
31589 (gimple_fold_builtin_snprintf): Same.
31590 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
31591 (gimple_fold_builtin_snprintf): Same.
31592 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
31593 of character types.
31594 (is_call_safe): New function.
31595 (try_substitute_return_value): Call it.
31596 (try_simplify_call): New function.
31597 (pass_sprintf_length::handle_gimple_call): Call it.
31599 2017-05-14 Martin Sebor <msebor@redhat.com>
31601 PR middle-end/80669
31602 * builtins.c (expand_builtin_stpncpy): Simplify.
31604 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
31606 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
31607 * config/i386/i386.h
31608 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
31609 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
31610 (struct machine_function): Add new members call_ms2sysv,
31611 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
31612 (struct machine_frame_state): New fields sp_realigned and
31613 sp_realigned_offset.
31614 * config/i386/i386.c
31615 (enum xlogue_stub): New enum.
31616 (enum xlogue_stub_sets): New enum.
31617 (class xlogue_layout): New class.
31618 (struct ix86_frame): New fields stack_realign_allocate_offset,
31619 stack_realign_offset and outlined_save_offset. Modify comments to
31620 detail stack layout when using out-of-line stubs.
31621 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
31622 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
31623 -mcall-ms2sysv-xlogues.
31624 (stub_managed_regs): New static variable.
31625 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
31626 registers managed by out-of-line stub.
31627 (disable_call_ms2sysv_xlogues): New function.
31628 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
31629 m->call_ms2sysv when appropriate and compute frame layout for
31631 (sp_valid_at, fp_valid_at): New inline functions.
31632 (choose_basereg): New function.
31633 (choose_baseaddr): Add align parameter, use choose_basereg and modify
31635 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
31636 Use align parameter of choose_baseaddr to generated aligned SSE movs
31638 (pro_epilogue_adjust_stack): Modify to track
31639 machine_frame_state::sp_realigned.
31640 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
31641 (ix86_nsaved_sseregs): Likewise.
31642 (ix86_emit_save_regs): Likewise.
31643 (ix86_emit_save_regs_using_mov): Likewise.
31644 (ix86_emit_save_sse_regs_using_mov): Likewise.
31645 (get_scratch_register_on_entry): Likewise.
31646 (gen_frame_set): New function.
31647 (gen_frame_load): Likewise.
31648 (gen_frame_store): Likewise.
31649 (emit_outlined_ms2sysv_save): Likewise.
31650 (emit_outlined_ms2sysv_restore): Likewise.
31651 (ix86_expand_prologue): Modify stack re-alignment code and call
31652 emit_outlined_ms2sysv_save when appropriate.
31653 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
31654 parameter rtx_insn *insn, which allows the function to be used to only
31655 generate the notes.
31656 (ix86_expand_epilogue): Modify validity checks of frame and stack
31657 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
31658 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
31659 * config/i386/predicates.md
31660 (save_multiple): New predicate.
31661 (restore_multiple): Likewise.
31662 * config/i386/sse.md
31663 (save_multiple<mode>): New pattern.
31664 (save_multiple_realign<mode>): Likewise.
31665 (restore_multiple<mode>): Likewise.
31666 (restore_multiple_and_return<mode>): Likewise.
31667 (restore_multiple_leave_return<mode>): Likewise.
31668 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
31670 2017-05-14 Julia Koval <julia.koval@intel.com>
31672 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
31673 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
31674 (__builtin_ia32_xsetbv): New builtins.
31675 * config/i386/i386.c (ix86_expand_special_args_builtin):
31677 (ix86_expand_builtin): Special expand for new intrinsics.
31678 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
31679 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
31680 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
31682 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31684 * cfganal.c (inverted_post_order_compute): Change argument type
31686 * cfganal.h (inverted_post_order_compute): Adjust prototype.
31687 * df-core.c (rest_of_handle_df_initialize): Adjust.
31688 (rest_of_handle_df_finish): Likewise.
31689 (df_analyze_1): Likewise.
31690 (df_analyze): Likewise.
31691 (loop_inverted_post_order_compute): Change argument to be a vec *.
31692 (df_analyze_loop): Adjust.
31693 (df_get_n_blocks): Likewise.
31694 (df_get_postorder): Likewise.
31695 * df.h (struct df_d): Change field to be a vec.
31696 * lcm.c (compute_laterin): Adjust.
31697 (compute_available): Likewise.
31698 * lra-lives.c (lra_create_live_ranges_1): Likewise.
31699 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
31700 * tree-ssa-pre.c (compute_antic): Likewise.
31702 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31704 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
31705 (depth_first_search::depth_first_search): Change structure init
31706 function to this constructor.
31707 (depth_first_search::add_bb): Rename function to this member.
31708 (depth_first_search::execute): Likewise.
31709 (flow_dfs_compute_reverse_finish): Adjust.
31711 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31713 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
31714 (longest_simple_path): Likewise.
31715 * shrink-wrap.c (spread_components): Likewise.
31716 (disqualify_problematic_components): Likewise.
31717 (emit_common_heads_for_components): Likewise.
31718 (emit_common_tails_for_components): Likewise.
31719 (insert_prologue_epilogue_for_components): Likewise.
31721 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31723 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
31726 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31728 * df-core.c (df_set_blocks): Start using auto_bitmap.
31729 (df_compact_blocks): Likewise.
31730 * df-problems.c (df_rd_confluence_n): Likewise.
31731 * df-scan.c (df_insn_rescan_all): Likewise.
31732 (df_process_deferred_rescans): Likewise.
31733 (df_update_entry_block_defs): Likewise.
31734 (df_update_exit_block_uses): Likewise.
31735 (df_entry_block_bitmap_verify): Likewise.
31736 (df_exit_block_bitmap_verify): Likewise.
31737 (df_scan_verify): Likewise.
31738 * lra-constraints.c (lra_constraints): Likewise.
31739 (undo_optional_reloads): Likewise.
31740 (lra_undo_inheritance): Likewise.
31741 * lra-remat.c (calculate_gen_cands): Likewise.
31742 (do_remat): Likewise.
31743 * lra-spills.c (assign_spill_hard_regs): Likewise.
31744 (spill_pseudos): Likewise.
31745 * tree-ssa-pre.c (bitmap_set_and): Likewise.
31746 (bitmap_set_subtract_values): Likewise.
31748 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31750 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
31751 management with auto_bitmap.
31752 (fix_inter_tick): Likewise.
31753 (fix_recovery_deps): Likewise.
31754 * ira.c (add_store_equivs): Likewise.
31755 (find_moveable_pseudos): Likewise.
31756 (split_live_ranges_for_shrink_wrap): Likewise.
31757 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
31758 (rtx_reuse_manager::seen_def_p): Likewise.
31759 (rtx_reuse_manager::set_seen_def): Likewise.
31760 * print-rtl.h (class rtx_reuse_manager): Likewise.
31762 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31764 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
31766 (migrate_btr_def): Likewise.
31767 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
31768 * df-core.c (loop_post_order_compute): Likewise.
31769 (loop_inverted_post_order_compute): Likewise.
31770 * hsa-common.h: Likewise.
31771 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
31772 * init-regs.c (initialize_uninitialized_regs): Likewise.
31773 * ipa-inline.c (resolve_noninline_speculation): Likewise.
31774 (inline_small_functions): Likewise.
31775 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
31776 * ira.c (combine_and_move_insns): Likewise.
31777 (build_insn_chain): Likewise.
31778 * loop-invariant.c (find_invariants): Likewise.
31779 * lower-subreg.c (propagate_pseudo_copies): Likewise.
31780 * predict.c (tree_predict_by_opcode): Likewise.
31781 (predict_paths_leading_to): Likewise.
31782 (predict_paths_leading_to_edge): Likewise.
31783 (estimate_loops_at_level): Likewise.
31784 (estimate_loops): Likewise.
31785 * shrink-wrap.c (try_shrink_wrapping): Likewise.
31786 (spread_components): Likewise.
31787 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
31788 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
31789 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
31790 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
31791 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
31792 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
31793 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
31794 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
31795 (mark_threaded_blocks): Likewise.
31796 (thread_through_all_blocks): Likewise.
31797 * tree-ssa.c (verify_ssa): Likewise.
31798 (execute_update_addresses_taken): Likewise.
31799 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
31801 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31803 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
31805 (post_order_compute): Likewise.
31806 (inverted_post_order_compute): Likewise.
31807 (pre_and_rev_post_order_compute_fn): Likewise.
31809 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31811 * genrecog.c (int_set::int_set): Explicitly construct our
31812 auto_vec base class.
31813 * vec.h (auto_vec::auto_vec): New constructor.
31815 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31817 * bitmap.h (class auto_bitmap): New constructor taking
31818 bitmap_obstack * argument.
31820 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
31822 * bitmap.h (class auto_bitmap): Change type of m_bits to
31823 bitmap_head, and adjust ctor / dtor and member operators.
31825 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
31827 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
31828 when returned register mode doesn't match original mode.
31830 2017-05-12 Jeff Law <law@redhat.com>
31831 Jakub Jelinek <jakub@redhat.com>
31833 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
31834 we look for cc setter after the compare-elim changes.
31835 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
31836 within the vector to match what compare-elim now expects.
31837 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
31838 (xorsi3_flags, one_cmplsi2_flags): Likewise.
31840 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
31841 after the compare-elim changes.
31842 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
31843 the vector to match what compare-elim now expects.
31844 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
31845 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
31846 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
31847 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
31848 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
31850 * config/visium/visium.c (single_set_and_flags): Fix where
31851 we look for cc setter after the compare-elim changes.
31852 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
31853 with the vector to match what compare-elim now expects.
31854 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
31855 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
31856 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
31857 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
31858 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
31859 (neg<mode>2_insn_set_overflow): Likewise.
31861 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
31863 PR middle-end/79794
31864 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
31865 maybe_expand_insn call, set ops[0].target. If still set after call,
31866 set alt_rtl. Add extra arg to recursive calls.
31867 (extract_bit_field): Add alt_rtl argument. Pass to
31869 * expmed.h (extract_bit_field): Fix prototype.
31870 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
31871 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
31872 to extract_bit_field_calls.
31873 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
31874 Pass alt_rtl to extract_bit_field calls.
31875 * calls.c (store_unaligned_arguments_into_psuedos)
31876 load_register_parameters): Pass extra NULL to extract_bit_field calls.
31877 * optabs.c (maybe_legitimize_operand): Clear op->target when call
31879 * optabs.h (struct expand_operand): Add target bitfield.
31881 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31883 * compare-elim.c (try_eliminate_compare): Canonicalize
31884 operation with embedded compare to
31885 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
31886 (set (reg) (operation)].
31888 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
31890 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
31893 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
31894 cost of adding a carry flag for ADC instruction.
31895 [case MINUS]: Ignore the cost of subtracting a carry flag
31896 for SBB instruction.
31898 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
31900 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
31902 * config/rs6000/bmiintrin.h: New file.
31903 * config/rs6000/bmi2intrin.h: New file.
31904 * config/rs6000/x86intrin.h: New file.
31906 2017-05-12 Jeff Law <law@redhat.com>
31908 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
31911 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
31913 PR middle-end/80707
31914 * tree-cfg.c: Remove cfg edges of unreachable case statements.
31916 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31918 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31919 early expansion of vector divide builtins.
31920 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
31921 builtins identified as having unsigned arguments.
31923 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31925 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
31926 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
31927 expansion of vector logical operations (and, andc, or, xor,
31930 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31932 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
31933 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
31935 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
31937 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
31938 early GIMPLE expansion of vector multiplies.
31940 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
31942 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
31943 TARGET_HAVE_MOVT conditional.
31944 (movt splitter): Likewise.
31946 2017-05-12 Richard Biener <rguenther@suse.de>
31948 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
31949 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
31950 Fold all stmts not inplace.
31952 2017-05-12 Richard Biener <rguenther@suse.de>
31954 PR tree-optimization/80713
31955 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
31956 inserted_exprs bit for not removed stmts.
31958 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
31960 PR middle-end/69921
31961 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
31962 parallelized" attribute for parallelized OpenACC kernels.
31963 * omp-offload.c (execute_oacc_device_lower): Use it.
31965 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
31966 Set "oacc kernels" attribute.
31967 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
31968 parameter. Adjust all users.
31969 (oacc_fn_attrib_kernels_p): Remove function.
31970 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
31971 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
31972 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
31973 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
31974 assert "oacc kernels" attribute is set.
31976 2017-05-11 Carl Love <cel@us.ibm.com>
31978 * config/rs6000/rs6000-c: Add support for built-in functions
31979 vector unsigned char vec_popcnt (vector signed char)
31980 vector unsigned char vec_popcnt (vector unsigned char)
31981 vector unsigned short vec_popcnt (vector signed short)
31982 vector unsigned short vec_popcnt (vector unsigned short)
31983 vector unsigned int vec_popcnt (vector signed int)
31984 vector unsigned int vec_popcnt (vector unsigned int)
31985 vector unsigned long long vec_popcnt (vector signed long long)
31986 vector unsigned long long vec_popcnt (vector unsigned long long)
31987 vector signed long long vec_slo (vector signed long long,
31988 vector signed char)
31989 vector signed long long vec_slo (vector signed long long,
31990 vector unsigned char)
31991 vector unsigned long long vec_slo (vector unsigned long long,
31992 vector signed char)
31993 vector unsigned long long vec_slo (vector unsigned long long,
31994 vector unsigned char)
31995 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
31996 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
31997 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
31998 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
31999 * doc/extend.texi: Update the built-in documentation file for the
32000 new built-in functions.
32002 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
32004 * attribs.h (sorted_attr_string): Move machine independent
32005 functions for target clone support from the i386 port to common
32006 code. Rename ix86_function_versions to common_function_versions.
32007 Rename make_name to make_unique_name.
32008 (common_function_versions): Likewise.
32009 (make_unique_name): Likewise.
32010 (make_dispatcher_decl): Likewise.
32011 (is_function_default_version): Likewise.
32012 * attribs.c (attr_strcmp): Likewise.
32013 (sorted_attr_string): Likewise.
32014 (common_function_versions): Likewise.
32015 (make_unique_name): Likewise.
32016 (make_dispatcher_decl): Likewise.
32017 (is_function_default_version): Likewise.
32018 * config/i386/i386.c (attr_strcmp): Likewise.
32019 (sorted_attr_string): Likewise.
32020 (ix86_function_versions): Likewise.
32021 (make_name): Likewise.
32022 (make_dispatcher_decl): Likewise.
32023 (is_function_default_version): Likewise.
32024 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
32026 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32029 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
32030 Account for direct move costs for vec_construct of integer
32033 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
32036 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
32037 (UNSPEC_STX_ATOMIC): Ditto.
32038 (loaddi_via_sse): New insn.
32039 (storedi_via_sse): Ditto.
32040 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
32041 Update corresponding peephole2 patterns.
32042 (atomic_storedi_fpu): Ditto.
32044 2017-05-11 Julia Koval <julia.koval@intel.com>
32046 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
32047 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
32049 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
32050 (__builtin_ia32_rsqrt14ss_mask): New builtins.
32051 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
32053 2017-05-11 Nathan Sidwell <nathan@acm.org>
32055 * graphite-poly.c: Include dumpfile.h.
32057 * dumpfle.h (dump_function): Declare here ...
32058 * tree-dump.h (dump_function): ... not here.
32059 * dumpfile.c: #include tree-cfg.h.
32060 (dump_function): Move here from ...
32061 * tree-dump.c (dump_function): ... here.
32062 * gimplify.c: #include splay-tree.h, not tree-dump.h.
32063 * graphite-poly.c: Don't include tree-dump.h.
32064 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
32065 * print-tree.c: Likewise.
32066 * stor-layout.c: Likewise.
32067 * tree-nested.c: Likewise.
32069 * dumpfile.c (dump_start): Use TDF_FLAGS.
32070 (dump_enable_all): Fix TDF_KIND check thinko.
32072 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
32074 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32075 array entries to represent two legal parameterizations of the
32076 overloaded __builtin_cmpb function, as represented by the
32077 P6_OV_BUILTIN_CMPB constant.
32078 (altivec_resolve_overloaded_builtin): Add special case handling
32079 for the __builtin_cmpb function, as represented by the
32080 P6_OV_BUILTIN_CMPB constant.
32081 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
32082 (BU_P6_64BIT_2): New macro.
32083 (BU_P6_OVERLOAD_2): New macro
32084 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
32085 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
32086 (CMPB): Add overload support to represent both 32-bit and 64-bit
32087 compare-bytes function.
32088 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
32089 support for TARGET_CMPB.
32090 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
32091 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
32092 documentation of the __builtin_cmpb overloaded built-in function.
32094 2017-05-11 Richard Biener <rguenther@suse.de>
32096 PR tree-optimization/80705
32097 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
32098 bases are not vectorizable.
32100 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32102 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
32103 when counting register pressure.
32105 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32107 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
32108 (iv_ca_more_deps): Renamed to ...
32109 (iv_ca_compare_deps): ... this.
32110 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
32112 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32114 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
32116 (determine_group_iv_costs): ... here.
32117 (find_inv_vars_cb): Record inv var if it's not recorded before.
32119 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32121 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
32122 (get_shiftadd_cost): Ditto.
32124 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32126 * tree-ssa-address.c: Include header file.
32127 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
32129 (add_to_parts): Refactor.
32130 (addr_to_parts): New parameter. Update use of move_hint_to_base.
32131 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
32134 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32136 PR tree-optimization/53090
32137 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
32139 (extract_cond_operands): Detect condition with IV on both sides
32140 and return COMP_IV_EXPR_2.
32141 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
32142 (rewrite_use_compare): Simplify by removing call to function
32143 extract_cond_operands.
32145 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32147 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
32148 (extract_cond_operands): Detect condition comparing against non-
32149 invariant bound and return appropriate enum value.
32150 (find_interesting_uses_cond): Update use of extract_cond_operands.
32151 Handle its return value accordingly.
32152 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
32154 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32156 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
32157 nonlinear iv_use computation in loop invariant sensitive way.
32159 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32161 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
32162 (find_iv_candidates): Call relate_compare_use_with_all_cands.
32164 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32166 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
32167 (dump_cand): Support iv_cand.inv_exprs.
32168 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
32170 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
32173 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32175 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
32177 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
32178 as local function. Include necessary header files.
32179 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
32181 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32183 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
32185 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32187 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
32188 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
32189 RSHIFT_EXPR and BIT_NOT_EXPR.
32191 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32193 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
32194 (adjust_setup_cost): New parameter supporting round up adjustment.
32195 (struct address_cost_data): Delete.
32196 (force_expr_to_var_cost): Don't bound cost with spill_cost.
32197 (split_address_cost, ptr_difference_cost): Delete.
32198 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
32199 (struct ainc_cost_data): New struct.
32200 (get_address_cost_ainc): New function.
32201 (get_address_cost, get_computation_cost): Reimplement.
32202 (determine_group_iv_cost_address): Record inv_expr for all uses of
32204 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
32205 (iv_ca_has_deps): Reimplemented to ...
32206 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
32208 (iv_ca_extend): Call iv_ca_more_deps.
32210 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32212 * tree-ssa-address.c (struct mem_address): Move to header file.
32213 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
32214 * tree-ssa-address.h (struct mem_address): Move from C file.
32215 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
32217 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32219 * tree-affine.h (aff_combination_type): New interface.
32220 (aff_combination_zero_p): Remove static.
32221 (aff_combination_const_p): New interface.
32222 (aff_combination_singleton_var_p): New interfaces.
32224 2017-05-11 Richard Biener <rguenther@suse.de>
32226 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
32227 Skip unreachable blocks and destinations.
32228 (eliminate): Move stmt removal and fixup ...
32229 (fini_eliminate): ... here. Skip inserted exprs.
32230 (pass_pre::execute): Move fini_pre after fini_eliminate.
32231 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
32232 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
32233 PRE to get rid of dead code that has invalid SSA form and
32234 split critical edges again.
32236 2017-05-11 Bin Cheng <bin.cheng@arm.com>
32238 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
32240 2017-05-11 Richard Biener <rguenther@suse.de>
32242 * passes.c (execute_function_todo): Verify loops if they are
32243 said to be up-to-date.
32244 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
32245 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
32247 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
32250 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
32251 handle calling assemble_external ourself.
32254 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
32255 modes with zero size. Enhance comment.
32257 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32259 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
32260 built-ins for vec_xl and vec_xst with short and char pointer
32263 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32265 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
32266 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
32267 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
32268 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
32269 (_mm_maskz_min_round_ss): New intrinsics.
32270 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
32271 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
32272 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
32273 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
32274 (__builtin_ia32_minss_mask_round): New builtins.
32275 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32276 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32277 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
32279 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
32280 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
32282 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
32285 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
32287 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
32288 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
32289 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
32290 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
32291 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
32292 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
32293 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
32294 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32295 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
32296 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
32297 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
32298 (__builtin_ia32_mulss_mask_round): New builtins.
32299 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
32300 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
32301 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
32303 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
32304 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
32306 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
32309 2017-05-10 Julia Koval <julia.koval@intel.com>
32311 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
32312 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
32313 (_mm256_setr_m128i): New intrinsics.
32315 2017-05-10 Julia Koval <julia.koval@intel.com>
32317 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
32318 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
32319 (_mm_maskz_rcp14_ss): New intrinsics.
32320 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
32321 (__builtin_ia32_rcp14ss_mask): New builtins.
32322 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
32324 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
32326 PR tree-optimization/51513
32327 * tree-cfg.c (gimple_seq_unreachable_p): New function.
32328 (assert_unreachable_fallthru_edge_p): Use it.
32329 (group_case_labels_stmt): Likewise.
32330 * tree-cfg.h: Prototype it.
32331 * stmt.c: Include cfghooks.h and tree-cfg.h.
32332 (emit_case_dispatch_table) <gap_label>: New local variable.
32333 Use it to fill dispatch table gaps.
32334 Test for default_label before updating probabilities.
32335 (expand_case) <default_label>: Remove unneeded initialization.
32336 Test for unreachable default case statement and remove its edge.
32337 Set default_label accordingly.
32338 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
32340 2017-05-10 Carl Love <cel@us.ibm.com>
32342 * config/rs6000/rs6000-c: Add support for built-in functions
32343 vector signed char vec_neg (vector signed char)
32344 vector signed short int vec_neg (vector short int)
32345 vector signed int vec_neg (vector signed int)
32346 vector signed long long vec_neg (vector signed long long)
32347 vector float vec_neg (vector float)
32348 vector double vec_neg (vector double)
32349 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
32351 * config/rs6000/altivec.h: Add define for vec_neg
32352 * doc/extend.texi: Update the built-in documentation for the
32353 new built-in functions.
32355 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32357 PR tree-optimization/77644
32358 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
32360 2017-05-10 Nathan Sidwell <nathan@acm.org>
32362 * dumpfile.h (TDI_lang_all): New.
32363 (TDF_KIND): New. Renumber others
32364 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
32366 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
32368 (get_dump_file_name): Adjust suffix generation.
32369 (dump_enable_all): Use TDF_KIND.
32370 * doc/invoke.texi (-fdump-lang-all): Document.
32372 * dumpfile.h: Tabify.
32374 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
32377 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
32378 Move member access before delete.
32380 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
32382 * tree-inline.c (expand_call_inline): Split block at stmt
32385 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
32388 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
32389 are now unused after splitting mov{sf,sd}_hardfloat.
32390 (f32_lr2): Likewise.
32391 (f32_lm): Likewise.
32392 (f32_lm2): Likewise.
32393 (f32_li): Likewise.
32394 (f32_li2): Likewise.
32395 (f32_lv): Likewise.
32396 (f32_sr): Likewise.
32397 (f32_sr2): Likewise.
32398 (f32_sm): Likewise.
32399 (f32_sm2): Likewise.
32400 (f32_si): Likewise.
32401 (f32_si2): Likewise.
32402 (f32_sv): Likewise.
32403 (f32_dm): Likewise.
32404 (f32_vsx): Likewise.
32405 (f32_av): Likewise.
32406 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
32407 For movsf, order stores so the VSX stores occur before the GPR
32408 store which encourages the register allocator to use a traditional
32409 FPR instead of a GPR. For movsd, order the stores so that the GPR
32410 store comes before the VSX stores to allow the power6 to work.
32411 This is due to the power6 not having a 32-bit integer store
32412 instruction from a FPR.
32413 (movsf_hardfloat): Likewise.
32414 (movsd_hardfloat): Likewise.
32416 2017-05-09 Martin Sebor <msebor@redhat.com>
32418 PR translation/80280
32419 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
32422 PR translation/80280
32423 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
32424 data member added in r247778.
32425 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
32427 2017-05-09 Nathan Sidwell <nathan@acm.org>
32429 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
32431 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
32434 2017-05-09 Marek Polacek <polacek@redhat.com>
32436 * doc/invoke.texi: Fix typo.
32438 2017-05-09 Richard Biener <rguenther@suse.de>
32440 * tree-vrp.c (vrp_val_is_max): Adjust comment.
32441 (vrp_val_is_min): Likewise.
32442 (set_value_range_to_value): Likewise.
32443 (set_value_range_to_nonnegative): Likewise.
32444 (gimple_assign_nonzero_p): Likewise.
32445 (gimple_stmt_nonzero_p): Likewise.
32446 (vrp_int_const_binop): Likewise. Remove unreachable case.
32447 (adjust_range_with_scev): Adjust comments.
32448 (compare_range_with_value): Likewise.
32449 (extract_range_from_phi_node): Likewise.
32450 (test_for_singularity): Likewise.
32452 2017-05-09 Richard Biener <rguenther@suse.de>
32454 * tree-vrp.c (get_single_symbol): Add assert that we don't
32455 get overflowed constants as invariant part.
32456 (compare_values_warnv): Add comment before the TREE_NO_WARNING
32457 checks. Use wi::cmp instead of recursing for integer constants.
32458 (compare_values): Just ignore whether we assumed undefined
32459 overflow instead of failing the compare.
32460 (extract_range_for_var_from_comparison_expr): Add comment before the
32461 TREE_NO_WARNING sets.
32462 (test_for_singularity): Likewise.
32463 (extract_range_from_comparison): Do not disable optimization
32464 when we assumed undefined overflow.
32465 (extract_range_basic): Remove init of unused var.
32467 2017-05-09 Richard Biener <rguenther@suse.de>
32469 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
32470 (extract_range_from_multiplicative_op_1): Adjust.
32471 (extract_range_from_binary_expr_1): Use int_const_binop.
32473 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
32476 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
32477 rs6000_store_data_bypass_p in seven define_bypass directives and
32478 in several comments.
32479 * config/rs6000/rs6000-protos.h: Add prototype for
32480 rs6000_store_data_bypass_p function.
32481 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
32482 function implements slightly different (rs6000-specific) semantics
32483 than store_data_bypass_p, returning false rather than aborting
32484 with assertion error when arguments do not satisfy the
32485 requirements of store data bypass.
32486 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
32487 rs6000_store_data_bypass_p.
32489 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
32491 * config/xtensa/xtensa-protos.h
32492 (xtensa_initial_elimination_offset): New declaration.
32493 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
32494 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
32495 macro definition, add case for FRAME_POINTER_REGNUM when
32496 FRAME_GROWS_DOWNWARD.
32497 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
32498 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
32499 xtensa_initial_elimination_offset.
32501 2017-05-08 Nathan Sidwell <nathan@acm.org>
32503 * doc/invoke.texi: Alphabetize -fdump options.
32505 2017-05-08 Martin Sebor <msebor@redhat.com>
32507 PR translation/80280
32508 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
32510 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
32512 * target.def (compute_frame_layout): New optional target hook.
32513 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
32514 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
32515 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
32517 * reload1.c (verify_initial_elim_offsets): Likewise.
32518 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
32519 (use_simple_return_p): Call arm_compute_frame_layout if needed.
32520 (arm_get_frame_offsets): Split up into this ...
32521 (arm_compute_frame_layout): ... and this function.
32523 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
32525 * config/aarch64/constraints.md (Usa): New constraint.
32526 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
32528 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
32530 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
32531 with_multilib_list after it has been checked.
32533 2017-05-08 Richard Biener <rguenther@suse.de>
32535 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
32536 (bitmap_set_subtract_values): Likewise.
32538 2017-05-08 Richard Biener <rguenther@suse.de>
32540 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
32541 (gimple_assign_nonzero): ... this and remove strict_overflow_p
32543 (gimple_stmt_nonzero_warnv_p): Rename to ...
32544 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
32546 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
32547 (extract_range_basic): Adjust, do not disable propagation on
32548 strict overflow sensitive simplification.
32549 (vrp_visit_cond_stmt): Likewise.
32551 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
32553 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
32554 body size unconditionally.
32556 2017-05-07 Jeff Law <law@redhat.com>
32559 2017-05-06 Jeff Law <law@redhat.com>
32560 PR tree-optimization/78496
32561 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32564 PR tree-optimization/78496
32565 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32566 (simplify_stmt_using_ranges): Call it.
32567 (vrp_dom_walker::before_dom_children): Extract equivalences
32568 from an ASSERT_EXPR with an equality comparison against a
32571 2017-05-06 Jeff Law <law@redhat.com>
32573 PR tree-optimization/78496
32574 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
32577 PR tree-optimization/78496
32578 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
32579 (simplify_stmt_using_ranges): Call it.
32580 (vrp_dom_walker::before_dom_children): Extract equivalences
32581 from an ASSERT_EXPR with an equality comparison against a
32584 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32586 * lra-constraints.c (lra_copy_reg_equiv): New function.
32587 (split_reg): Use it to copy equivalence information from the
32588 original register to the spill register.
32590 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
32592 PR rtl-optimization/75964
32593 * simplify-rtx.c (simplify_const_relational_operation): Remove
32594 invalid handling of comparisons of integer ABS.
32596 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
32598 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
32599 initialize to zero.
32600 (init_regs): Remove declaration.
32601 (function_arg_advance_32): Initialize error_p as boolean variable.
32603 2017-05-05 Nathan Sidwell <nathan@acm.org>
32605 * store-motion.c (remove_reachable_equiv_notes): Reformat long
32606 lines. Use for (;;).
32608 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32610 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
32611 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
32612 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
32613 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
32614 VF=2 that require versioning.
32616 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32618 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
32621 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32623 * diagnostic.h (diagnostic_override_option_index): Convert from
32624 macro to inline function.
32626 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32628 * diagnostic.c (last_module_changed_p): New function.
32629 (set_last_module): New function.
32630 (diagnostic_report_current_module): Convert macro usage to
32631 the above functions.
32632 * diagnostic.h (diagnostic_context::last_module): Strengthen
32633 from const line_map * to const line_map_ordinary *.
32634 (diagnostic_last_module_changed): Delete macro.
32635 (diagnostic_set_last_module): Delete macro.
32637 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32639 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
32640 with diagnostic_report_diagnostic.
32641 (diagnostic_n_impl_richloc): Likewise.
32642 * diagnostic.h (report_diagnostic): Delete macro.
32643 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
32644 with diagnostic_report_diagnostic.
32645 * substring-locations.c (format_warning_va): Likewise.
32647 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32649 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
32650 save/restor of format_spec. Move option-printing code to...
32651 (print_option_information): ...this new function, and
32652 reimplement by simply printing to the pretty_printer,
32653 rather than appending to the format string.
32655 2017-05-05 David Malcolm <dmalcolm@redhat.com>
32657 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
32658 handling logic into...
32659 (update_effective_level_from_pragmas): ...this new function.
32661 2017-05-04 Andrew Waterman <andrew@sifive.com>
32663 * config/riscv/riscv.opt (mstrict-align): New option.
32664 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
32665 (SLOW_UNALIGNED_ACCESS): Define.
32666 (riscv_slow_unaligned_access): Declare.
32667 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
32669 (riscv_slow_unaligned_access): New variable.
32670 (rocket_tune_info): Set slow_unaligned_access to true.
32671 (optimize_size_tune_info): Set slow_unaligned_access to false.
32672 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
32673 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
32674 (riscv_option_override): Set riscv_slow_unaligned_access.
32675 * doc/invoke.texi: Add -mstrict-align to RISC-V.
32677 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
32679 * config/riscv/riscv.md: Unify indentation.
32681 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
32686 * config/rs6000/rs6000.md (u code attribute): Add FIX and
32688 (extendsi<mode>2): Add support for doing sign extension via
32689 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
32690 don't have ISA 3.0 instructions.
32691 (extendsi<mode>2 splitter): Likewise.
32692 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
32693 generate the normal insns since SImode can now go in vector
32694 registers. Disallow the special UNSPECs needed for previous
32695 machines to hide SImode being used. Add new insns
32696 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
32697 (fix_trunc<mode>si2_stfiwx): Likewise.
32698 (fix_trunc<mode>si2_internal): Likewise.
32699 (fixuns_trunc<mode>si2): Likewise.
32700 (fixuns_trunc<mode>si2_stfiwx): Likewise.
32701 (fctiw<u>z_<mode>_smallint): Likewise.
32702 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
32703 of floating point to 32-bit integer from doing a direct move to
32704 the GPR registers to do a store.
32705 (fctiwz_<mode>): Break long line.
32707 2017-05-05 Bin Cheng <bin.cheng@arm.com>
32709 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
32710 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
32711 (addr_list, addr_offset_valid_p): New.
32712 (split_address_groups): Check offset validity with above function.
32713 (gt-tree-ssa-loop-ivopts.h): Include header file.
32715 2017-05-05 Nathan Sidwell <nathan@acm.org>
32717 * config.gcc (arm*-*-*): Add missing 'fi'.
32719 2017-05-05 Steve Ellcey <sellcey@cavium.com>
32721 * doc/invoke.texi (-fopt-info): Explicitly say order of options
32722 included in -fopt-info does not matter.
32723 * doc/optinfo.texi (-fopt-info): Fix description of default
32724 behavour. Explicitly say order of options included in -fopt-info
32727 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
32729 * config.gcc: Allow combinations of aprofile and rmprofile values for
32730 --with-multilib-list.
32731 * config/arm/t-multilib: New file.
32732 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
32733 variables. Remove setting of ISA and floating-point ABI in
32734 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
32735 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
32736 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
32737 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
32739 * config/arm/t-rmprofile: Likewise except for the matches changes.
32740 * doc/install.texi (--with-multilib-list): Document the combination of
32741 aprofile and rmprofile values and warn about pitfalls in doing that.
32743 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32745 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
32746 (movdi_aarch64): Likewise.
32748 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32750 PR tree-optimization/80632
32751 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
32753 (build_arrays): Initialize it for virtual phis.
32754 (fix_phi_nodes): Use it for virtual phis.
32756 PR tree-optimization/80558
32757 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
32758 [x, y] op z into [x op, y op z] for op & or | if conditions
32761 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
32762 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32765 * config/arm/arm.md (use_literal_pool): Remove.
32766 (64-bit immediate split): No longer takes cost into consideration
32767 if arm_disable_literal_pool is enabled.
32768 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
32769 used when arm_disable_literal_pool is enabled.
32770 (arm_max_const_double_inline_cost): Remove use of
32771 arm_disable_literal_pool.
32772 (push_minipool_fix): Add assert.
32773 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
32774 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
32775 (no_literal_pool_sf_immediate): New.
32777 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32779 PR tree-optimization/80613
32780 * tree-ssa-dce.c (propagate_necessity): Remove cases for
32781 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
32783 2017-05-05 Richard Biener <rguenther@suse.de>
32785 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
32787 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
32789 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
32790 of this flag from insn conditions due to removal from r247495.
32792 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
32794 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
32796 (arm_early_store_addr_dep_ptr): Likewise.
32797 * config/arm/aarch-common-protos.h
32798 (arm_early_load_addr_dep_ptr): Add prototype.
32799 (arm_early_store_addr_dep_ptr): Likewise.
32800 * config/arm/cortex-a53.md: Add new bypasses.
32802 2017-05-05 Jakub Jelinek <jakub@redhat.com>
32804 * tree.c (next_type_uid): Change type to unsigned.
32805 (type_hash_canon): Decrement back next_type_uid if
32806 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
32807 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
32810 2017-05-04 Martin Sebor <msebor@redhat.com>
32812 * builtins.c: Fix a trivial typo in a comment.
32814 PR middle-end/79234
32815 * builtins.c (check_sizes): Adjust to handle reading past the end.
32816 Avoid printing excessive upper bound of ranges. Use %E to print
32817 tree nodes instead of converting them to %wu.
32818 (expand_builtin_memchr): New function.
32819 (compute_dest_size): Rename...
32820 (compute_objsize): ...to this.
32821 (expand_builtin_memcpy): Adjust.
32822 (expand_builtin_mempcpy): Adjust.
32823 (expand_builtin_strcat): Adjust.
32824 (expand_builtin_strcpy): Adjust.
32825 (check_strncat_sizes): Adjust.
32826 (expand_builtin_strncat): Adjust.
32827 (expand_builtin_strncpy): Adjust and simplify.
32828 (expand_builtin_memset): Adjust.
32829 (expand_builtin_bzero): Adjust.
32830 (expand_builtin_memcmp): Adjust.
32831 (expand_builtin): Handle memcmp.
32832 (maybe_emit_chk_warning): Check strncat just once.
32834 2017-05-04 Martin Sebor <msebor@redhat.com>
32836 PR preprocessor/79214
32837 PR middle-end/79222
32838 PR middle-end/79223
32839 * builtins.c (check_sizes): Add inlining context and issue
32840 warnings even when -Wno-system-headers is set.
32841 (check_strncat_sizes): Same.
32842 (expand_builtin_strncat): Same.
32843 (expand_builtin_memmove): New function.
32844 (expand_builtin_stpncpy): Same.
32845 (expand_builtin): Handle memmove and stpncpy.
32847 2017-05-04 Bin Cheng <bin.cheng@arm.com>
32849 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
32850 which is not used any more.
32852 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32854 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
32856 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32858 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
32859 (cortexa53_tunings): Likewise.
32860 (cortexa57_tunings): Likewise.
32861 (cortexa72_tunings): Likewise.
32862 (cortexa73_tunings): Likewise.
32864 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32866 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
32867 Set loop alignment to 8.
32869 2017-05-04 Martin Sebor <msebor@redhat.com>
32871 PR translation/80280
32872 * builtins.c (expand_builtin_object_size): Add missing quoting to
32873 %D and like directives.
32874 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
32875 (hsa_type_for_tree_type): Same.
32876 (verify_function_arguments): Same.
32877 * symtab.c (symbol_table::change_decl_assembler_name): Same.
32878 * varasm.c (get_section): Same.
32881 2017-05-04 Martin Sebor <msebor@redhat.com>
32883 PR translation/80280
32884 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
32886 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
32888 * config/aarch64/aarch64.c (generic_addrcost_table):
32889 Change HI/TI mode setting.
32891 2017-05-04 Martin Jambor <mjambor@suse.cz>
32893 PR tree-optimization/80622
32894 * tree-sra.c (comes_initialized_p): New function.
32895 (build_accesses_from_assign): Only set write lazily when
32896 comes_initialized_p is false.
32897 (analyze_access_subtree): Use comes_initialized_p.
32898 (propagate_subaccesses_across_link): Assert !comes_initialized_p
32899 instead of testing for PARM_DECL.
32901 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32903 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
32904 constraint on operand 0 to allow more general addressing modes.
32905 Adjust output template.
32906 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
32908 * config/aarch64/aarch64-protos.h
32909 (aarch64_address_valid_for_prefetch_p): Declare prototype.
32910 * config/aarch64/constraints.md (Dp): New address constraint.
32911 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
32914 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
32916 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
32917 update use of estimate_ipcp_clone_size_and_time.
32918 (estimate_local_effects): Update use of
32919 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
32920 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
32921 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
32922 Return nonspecialized time.
32924 2017-05-04 Richard Biener <rguenther@suse.de>
32926 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
32927 for the last VUSE which def dominates the PHI. Directly call
32929 (get_continuation_for_phi_1): Remove.
32931 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
32933 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
32934 to explain the use of truncating division. Cap the number of
32935 iterations to the maximum given by nb_iterations_upper_bound,
32938 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
32940 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
32941 * configure: Regenerate.
32942 * config.in: Regenerate.
32943 * config/i386/driver-mingw32.c: new file.
32944 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
32945 * config.host: Link driver-mingw32.o on MinGW host.
32946 * doc/install.texi: Document new --enable-mingw-wildcard configure
32949 2017-05-04 Marek Polacek <polacek@redhat.com>
32951 PR tree-optimization/80612
32952 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
32954 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32955 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
32957 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
32958 (movt splitter): Likewise.
32959 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
32960 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
32961 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
32962 block for Thumb-1 with MOVT.
32963 (thumb2_legitimate_address_p): Move code block ...
32964 (can_avoid_literal_pool_for_label_p): ... into this new function.
32965 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
32967 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
32968 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
32969 "M-profile targets with the MOVT instruction".
32971 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
32973 * config/arm/arm-builtins.c (arm_init_builtins): Rename
32974 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
32975 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
32977 2017-05-04 Martin Liska <mliska@suse.cz>
32979 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
32980 variable cond_code.
32982 2017-05-04 Richard Biener <rguenther@suse.de>
32984 * tree.c (array_at_struct_end_p): Handle arrays at struct
32985 end with flexarrays more conservatively. Refactor and treat
32986 arrays of arrays or aggregates more strict. Fix
32987 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
32988 * tree.c (array_at_struct_end_p): Adjust prototype.
32989 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
32990 * gimple-fold.c (get_range_strlen): Likewise.
32991 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
32993 2017-05-04 Richard Biener <rguenther@suse.de>
32995 PR tree-optimization/31130
32996 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
32998 (supports_overflow_infinity): Likewise.
32999 (is_negative_overflow_infinity): Likewise.
33000 (is_positive_overflow_infinity): Likewise.
33001 (is_overflow_infinity): Likewise.
33002 (stmt_overflow_infinity): Likewise.
33003 (overflow_infinity_range_p): Likewise.
33004 (usable_range_p): Remove as always returning true.
33005 (make_overflow_infinity): Remove.
33006 (negative_overflow_infinity): Likewise.
33007 (positive_overflow_infinity): Likewise.
33008 (avoid_overflow_infinity): Likewise.
33009 (set_value_range): Adjust accordingly.
33010 (set_value_range_to_nonnegative): Likewise, remove now unused
33011 overflow_infinity arg.
33012 (vrp_operand_equal_p): Adjust.
33013 (update_value_range): Likewise.
33014 (range_int_cst_singleton_p): Likewise.
33015 (operand_less_p): Likewise.
33016 (compare_values_warnv): Likewise.
33017 (extract_range_for_var_from_comparison_expr): Likewise.
33018 (vrp_int_const_binop): Likewise.
33019 (zero_nonzero_bits_from_vr): Likewise.
33020 (extract_range_from_multiplicative_op_1): Likewise.
33021 (extract_range_from_binary_expr_1): Likewise.
33022 (extract_range_from_unary_expr): Likewise.
33023 (extract_range_from_comparison): Likewise.
33024 (extract_range_basic): Likewise.
33025 (adjust_range_with_scev): Likewise.
33026 (compare_ranges): Likewise.
33027 (compare_range_with_value): Likewise.
33028 (dump_value_range): Likewise.
33029 (test_for_singularity): Likewise, remove strict_overflow_p parameter
33031 (simplify_cond_using_ranges): Adjust.
33033 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
33035 * brig-builtins.def: Added a builtin for class_f64.
33036 * builtin-types.def: Added a builtin type needed by class_f64.
33038 2017-05-03 Jason Merrill <jason@redhat.com>
33040 * timevar.def: Add TV_CONSTEXPR.
33042 2017-05-03 David Malcolm <dmalcolm@redhat.com>
33044 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
33046 2017-05-03 Martin Jambor <mjambor@suse.cz>
33048 * ipa-prop.c (ipa_update_after_lto_read): Removed.
33049 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
33050 * ipa-cp.c (ipcp_propagate_stage): Do not call
33051 ipa_update_after_lto_read.
33052 * ipa-inline.c (ipa_inline): Likewise.
33054 2017-05-03 Martin Jambor <mjambor@suse.cz>
33056 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
33057 tag. Added a default constructor and a destructor.
33058 (ipa_edge_args_sum_t): New class;
33059 (ipa_edge_args_sum): Declare.
33060 (ipa_edge_args_vector): Remove declaration.
33061 (IPA_EDGE_REF): Use ipa_edge_args_sum.
33062 (ipa_free_edge_args_substructures): Remove declaration.
33063 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
33064 (ipa_edge_args_info_available_for_edge_p): Likewise.
33065 * ipa-prop.c (ipa_edge_args_vector): Removed.
33066 (edge_removal_hook_holder): Likewise.
33067 (edge_duplication_hook_holder): Likewise.
33068 (ipa_edge_args_sum): New variable.
33069 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
33070 ipa_edge_args_vector.
33071 (ipa_free_edge_args_substructures): Likewise.
33072 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
33073 ipa_edge_args_vector.
33074 (ipa_edge_removal_hook): Turned into method
33075 ipa_edge_args_sum_t::remove.
33076 (ipa_edge_duplication_hook): Turned into method
33077 ipa_edge_args_sum_t::duplicate.
33078 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
33079 registering edge hooks.
33080 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
33081 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
33082 ipa_edge_args_sum instead of ipa_edge_args_vector.
33083 * ipa-profile.c (ipa_profile): Likewise.
33085 2017-05-03 Martin Jambor <mjambor@suse.cz>
33087 * symbol-summary.h (function_summary): New method exists.
33088 (function_summary::symtab_removal): Deallocate through release.
33089 (call_summary): New class.
33090 (gt_ggc_mx): New overload.
33091 (gt_pch_nx): Likewise.
33092 (gt_pch_nx): Likewise.
33094 2017-05-03 Jeff Law <law@redhat.com>
33096 PR tree-optimization/78496
33097 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
33098 from simplify_cond_using_ranges. Split off code to walk
33099 backwards through casts into ...
33100 (simplify_cond_using_ranges_2): New function.
33101 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
33102 (execute_vrp): After identifying jump threads, call
33103 simplify_cond_using_ranges_2.
33105 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
33108 * ipa-inline.h (inline_summary): Add ctor.
33109 (create_ggc): Do not use ggc_cleared_alloc.
33111 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
33112 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
33114 * gcc.c (handle_braces): Support escaping in switch matching
33116 * doc/invoke.texi (Spec Files): Document it.
33117 Remove superfluous @code markup in items.
33119 2017-05-03 David Malcolm <dmalcolm@redhat.com>
33121 * diagnostic-show-locus.c (struct column_range): New struct.
33122 (get_affected_columns): New function.
33123 (get_printed_columns): New function.
33124 (struct correction): New struct.
33125 (correction::ensure_capacity): New function.
33126 (correction::ensure_terminated): New function.
33127 (struct line_corrections): New struct.
33128 (line_corrections::~line_corrections): New dtor.
33129 (line_corrections::add_hint): New function.
33130 (layout::print_trailing_fixits): Reimplement in terms of the new
33132 (selftest::test_overlapped_fixit_printing): New function.
33133 (selftest::diagnostic_show_locus_c_tests): Call it.
33135 2017-05-03 Nathan Sidwell <nathan@acm.org>
33137 Canonicalize canonical type hashing
33138 * tree.h (type_hash_canon_hash): Declare.
33139 * tree.c (type_hash_list, attribute_hash_list): Move into
33140 type_hash_canon_hash.
33141 (build_type_attribute_qual_variant): Break out hash code calc into
33142 type_hash_canon_hash.
33143 (type_hash_canon_hash): New. Generic type hash computation.
33144 (build_range_type_1, build_array_type_1, build_function_type,
33145 build_method_type_directly, build_offset_type, build_complex_type,
33146 make_vector_type): Call it.
33148 2017-05-03 Richard Biener <rguenther@suse.de>
33150 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
33151 When all DRs have unknown misaligned do not always peel
33152 when there is a store but apply the same costing model as if
33153 there were only loads.
33155 2017-05-03 Richard Biener <rguenther@suse.de>
33158 PR tree-optimization/80492
33159 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33160 compare_base_decls returning dont-know properly.
33162 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
33164 * config/arm/iterators.md (CCSI): New mode iterator.
33165 (arch): New mode attribute.
33166 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
33167 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
33168 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
33169 code iterator for success result mode.
33170 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
33171 the corresponding new insn generators.
33173 2017-05-03 Bin Cheng <bin.cheng@arm.com>
33176 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33177 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33179 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
33181 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
33182 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
33183 (DDR_A): Wrap DDR argument in brackets.
33184 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
33185 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
33186 (DDR_REVERSED_P): Likewise.
33188 2017-05-03 Jakub Jelinek <jakub@redhat.com>
33190 PR tree-optimization/79472
33191 * tree-switch-conversion.c (struct switch_conv_info): Add
33192 contiguous_range and default_case_nonstandard fields.
33193 (collect_switch_conv_info): Compute contiguous_range and
33194 default_case_nonstandard fields, don't clear final_bb if
33195 contiguous_range and only the default case doesn't have the required
33197 (check_all_empty_except_final): Set default_case_nonstandard instead
33198 of failing if contiguous_range and the default case doesn't have empty
33200 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
33201 and only the default case doesn't have the required constants. Skip
33203 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
33204 if default_case_nonstandard.
33205 (build_constructors): Build constant 1 just once. Assert that default
33206 values aren't inserted in between cases if contiguous_range. Skip
33208 (build_arrays): Skip virtual phis.
33209 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
33210 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
33211 Handle virtual phis.
33212 (gen_inbound_check): Handle default_case_nonstandard case.
33213 (process_switch): Adjust check_final_bb caller. Call
33214 gather_default_values with the first non-default case instead of
33215 default case if default_case_nonstandard.
33217 2017-05-02 Nathan Sidwell <nathan@acm.org>
33219 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
33220 check. Fix formatting.
33222 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
33224 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
33225 errors when comparing specialized and unspecialized times.
33227 2017-05-02 David Malcolm <dmalcolm@redhat.com>
33229 * diagnostic-show-locus.c
33230 (layout::should_print_annotation_line_p): Make private.
33231 (layout::print_annotation_line): Make private.
33232 (layout::annotation_line_showed_range_p): Make private.
33233 (layout::show_ruler): Make private.
33234 (layout::print_source_line): Make private. Pass in line and
33235 line_width, rather than calling location_get_source_line. Drop
33237 (layout::print_leading_fixits): New method.
33238 (layout::print_any_fixits): Rename to...
33239 (layout::print_trailing_fixits): ...this, and make private.
33240 Don't print newline fixits.
33241 (diagnostic_show_locus): Move logic for printing one row into...
33242 (layout::print_line): ...this new function. Move the
33243 location_get_source_line call and error-handling from
33244 print_source_line to here. Call print_leading_fixits, and rename
33245 print_any_fixits to print_trailing_fixits.
33246 (selftest::test_fixit_insert_containing_newline): Update now that
33247 newlines are partially supported.
33248 (selftest::test_fixit_insert_containing_newline_2): New test.
33249 (selftest::test_fixit_replace_containing_newline): Update comments.
33250 (selftest::diagnostic_show_locus_c_tests): Call the new test.
33251 * edit-context.c (class added_line): New class.
33252 (class edited_line): Describe newline handling in comment.
33253 (edited_line::actually_edited_p): New method.
33254 (edited_line::print_content): Delete redundant decl.
33255 (edited_line::m_predecessors): New field.
33256 (edited_file::print_content): Call edited_line::print_content.
33257 (edited_file::print_diff): Update to support newlines.
33258 (edited_file::print_diff_hunk): Likewise.
33259 (edited_file::print_run_of_changed_lines): New function.
33260 (edited_file::print_diff_line): Convert to...
33261 (print_diff_line): ...this.
33262 (edited_file::get_effective_line_count): New function.
33263 (edited_line::edited_line): Initialize new field m_predecessors.
33264 (edited_line::~edited_line): Clean up m_predecessors.
33265 (edited_line::apply_fixit): Handle newlines.
33266 (edited_line::get_effective_line_count): New function.
33267 (edited_line::print_content): New function.
33268 (edited_line::print_diff_lines): New function.
33269 (selftest::test_applying_fixits_insert_containing_newline): New
33271 (selftest::test_applying_fixits_replace_containing_newline): New
33273 (selftest::insert_line): New function.
33274 (selftest::test_applying_fixits_multiple_lines): Add example of
33276 (selftest::edit_context_c_tests): Call the new tests.
33278 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33280 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
33281 parameter cand. Update dump information.
33282 (get_computation_cost): Update uses.
33284 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33286 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
33287 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
33288 (get_computation_at, rewrite_use_address): Update use of
33289 get_computation_aff.
33291 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33293 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
33294 (get_computation): Delete.
33295 (get_computation_cost): Implement like get_computation_cost_at.
33296 Use get_computation_at.
33297 (get_computation_cost_at): Delete.
33298 (rewrite_use_nonlinear_expr): Use get_computation_at.
33299 (rewrite_use_compare, remove_unused_ivs): Ditto.
33301 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33303 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
33305 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33307 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
33308 (ivopts_global_cost_for_size): Rename parameter and update uses.
33309 (iv_ca_recount_cost): Update uses.
33310 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
33311 candidates seperately in n_invs and n_cands.
33312 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
33314 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33316 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
33317 (find_inv_vars_cb): New.
33318 (find_depends): Renamed to ...
33319 (find_inv_vars): ... this.
33320 (add_candidate_1, force_var_cost): Call find_inv_vars.
33321 (split_address_cost, determine_group_iv_cost_cond): Ditto.
33323 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33325 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
33326 inv_vars. Add inv_exprs.
33327 (struct iv_cand): Rename depends_on to inv_vars.
33328 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
33329 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
33330 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
33331 (dump_cand): Dump inv_vars.
33332 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
33333 (record_invariant, find_depends, add_candidate_1): Ditto.
33334 (set_group_iv_cost, force_var_cost): Ditto.
33335 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
33336 (get_computation_cost_at, get_computation_cost): Ditto.
33337 (determine_group_iv_cost_generic): Ditto.
33338 (determine_group_iv_cost_address): Ditto.
33339 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
33340 (determine_group_iv_costs): Ditto.
33341 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
33342 (iv_ca_set_remove_invariants): Renamed to ...
33343 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
33344 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
33345 (iv_ca_set_add_invariants): Renamed to ...
33346 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
33347 (iv_ca_set_cp): Use iv_ca_set_add_invs.
33348 (iv_ca_has_deps): Support inv_vars and inv_exprs.
33349 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
33350 (create_new_ivs): Remove useless dump.
33352 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33354 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
33356 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
33357 (iv_ca_set_no_cp, create_new_iv): Ditto.
33359 2017-05-02 Bin Cheng <bin.cheng@arm.com>
33361 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
33363 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33365 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
33366 function tree_check2.
33368 2017-05-02 Martin Liska <mliska@suse.cz>
33370 * doc/gcov.texi: Add missing preposition.
33371 * gcov.c (function_info::function_info): Properly fill up
33372 all member variables.
33374 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33376 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
33378 2017-05-02 Tamar Christina <tamar.christina@arm.com>
33380 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
33382 2017-05-02 Martin Liska <mliska@suse.cz>
33385 * lto-streamer-in.c (lto_read_tree_1): Remove
33386 LTO_STREAMER_DEBUG.
33387 * lto-streamer.c (struct tree_hash_entry): Likewise.
33388 (struct tree_entry_hasher): Likewise.
33389 (tree_entry_hasher::hash): Likewise.
33390 (tree_entry_hasher::equal): Likewise.
33391 (lto_streamer_init): Likewise.
33392 (lto_orig_address_map): Likewise.
33393 (lto_orig_address_get): Likewise.
33394 (lto_orig_address_remove): Likewise.
33395 * lto-streamer.h: Likewise.
33396 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
33397 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
33399 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
33401 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
33402 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
33403 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
33404 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
33405 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
33406 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
33407 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
33408 (mm_maskz_sub_ss): New intrinsics.
33409 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33410 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
33411 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
33412 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
33413 (__builtin_ia32_subss_mask_round): New builtins.
33414 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
33415 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
33416 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
33418 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
33419 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
33421 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
33424 2017-05-02 Martin Jambor <mjambor@suse.cz>
33426 PR tree-optimization/78687
33427 * tree-sra.c (access): New field parent.
33428 (process_subtree_disqualification): New function.
33429 (disqualify_candidate): Call it.
33430 (build_accesses_from_assign): Reset write flag if creating an
33432 (build_access_subtree): Fill in parent field and also prpagate
33433 down grp_write flag.
33434 (create_artificial_child_access): New parameter set_grp_write, set
33435 grp_write to its value.
33436 (propagate_subaccesses_across_link): Also propagate grp_write flag
33438 (propagate_all_subaccesses): Push the closest parent back to work
33439 queue if add_access_to_work_queue returned true.
33441 2017-05-02 Richard Biener <rguenther@suse.de>
33443 * common.opt (fstrict-overflow): Alias negative to fwrapv.
33444 * doc/invoke.texi (fstrict-overflow): Remove all traces of
33445 -fstrict-overflow documentation.
33446 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
33447 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
33448 flag_strict_overflow.
33449 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
33450 * lto-opts.c (lto_write_options): Do not stream it.
33451 * lto-wrapper.c (merge_and_complain): Do not handle it.
33452 * opts.c (default_options_table): Do not set -fstrict-overflow.
33453 (finish_options): Likewise do not clear it when sanitizing.
33454 * simplify-rtx.c (simplify_const_relational_operation): Do not
33455 test flag_strict_overflow.
33457 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33459 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
33460 using enabled attribute.
33461 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
33462 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
33463 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
33464 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
33465 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
33466 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
33467 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
33468 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
33469 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
33470 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
33472 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
33474 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
33476 2017-05-02 Richard Biener <rguenther@suse.de>
33478 PR tree-optimization/80591
33480 2017-04-10 Richard Biener <rguenther@suse.de>
33482 * tree-ssa-structalias.c (find_func_aliases): Properly handle
33485 2017-05-02 Richard Biener <rguenther@suse.de>
33487 PR tree-optimization/80549
33488 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
33489 (cleanup_tree_cfg_noloop): Create forwarders to known loop
33490 headers if they do not have a preheader.
33492 2017-05-02 Martin Liska <mliska@suse.cz>
33495 * common.opt: Fix typo.
33496 * doc/invoke.texi: Likewise.
33498 2017-05-01 Jan Beulich <jbeulich@suse.com>
33500 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
33501 swapping, add (x,x,m,x,n) alternative.
33503 2017-05-01 Nathan Sidwell <nathan@acm.org>
33505 * calls.c (combine_pending_stack_adjustment_and_call): Remove
33506 unnecessary unadjusted_alignment check.
33508 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
33511 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
33513 * gimplify.c (gimplify_cilk_detach): New function.
33514 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
33515 * tree-core.h: Document EXPR_CILK_SPAWN.
33516 * tree.h (EXPR_CILK_SPAWN): Define.
33518 2017-05-01 David Malcolm <dmalcolm@redhat.com>
33520 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
33521 to use new fixit_hint representation, using the "replace" logic.
33522 (get_line_span_for_fixit_hint): Likewise.
33523 (layout::print_any_fixits): Likewise.
33524 (selftest::test_one_liner_many_fixits): Rename to...
33525 (selftest::test_one_liner_many_fixits_1): ...this, and update
33526 comment and expected output to reflect that the multiple fix-it
33527 hints are now consolidated into one insertion.
33528 (selftest::test_one_liner_many_fixits_2): New test.
33529 (selftest::test_diagnostic_show_locus_one_liner): Update for
33531 (selftest::test_fixit_consolidation): Update for fix-it API
33533 * diagnostic.c (print_parseable_fixits): Likewise.
33534 * edit-context.c (edited_line::m_line_events): Convert from
33535 auto_vec <line_event *> to auto_vec <line_event>.
33536 (class line_event): Convert from abstract base class to a concrete
33537 class, taking over the role of replace_event.
33538 (class insert_event): Delete.
33539 (class replace_event): Rename to class line_event. Convert to
33541 (edit_context::add_fixits): Reimplement.
33542 (edit_context::apply_insert): Delete.
33543 (edit_context::apply_replace): Rename to...
33544 (edit_context::apply_fixit): ...this. Convert to half-open range.
33545 (edited_file::apply_insert): Delete.
33546 (edited_file::apply_replace): Rename to...
33547 (edited_file::apply_fixit): ...this.
33548 (edited_line::~edited_line): Drop deletion of events.
33549 (edited_line::apply_insert): Delete.
33550 (edited_line::apply_replace): Rename to...
33551 (edited_line::apply_fixit): ...this. Convert to half-open range.
33552 Update for change to type of m_line_events.
33553 * edit-context.h (edit_context::apply_insert): Delete.
33554 (edit_context::apply_replace): Rename to...
33555 (edit_context::apply_fixit): ...this.
33557 2017-05-01 Martin Sebor <msebor@redhat.com>
33559 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
33562 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
33565 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
33566 __get_cpuid_max returns 0.
33567 (__get_cpuid_count): Ditto.
33569 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
33571 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
33572 replacement expression is another instance of one of its arguments.
33574 2017-05-01 Jakub Jelinek <jakub@redhat.com>
33577 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
33578 check for stack push/pop autoinc.
33579 * config/i386/i386.c (ix86_agi_dependent): Return false
33580 if the only reason why modified_in_p returned true is that
33581 addr is SP based and set_insn is a push or pop.
33583 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33585 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
33588 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
33591 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
33592 (account_size_time): Use two predicates - exec_pred and
33594 (evaluate_conditions_for_known_args): Compute both clause and
33596 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
33597 (inline_summary_t::duplicate): Update.
33598 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
33600 (compute_inline_parameters): Likewise.
33601 (estimate_edge_size_and_time): Update caluclation of time.
33602 (estimate_node_size_and_time): Compute both time and nonspecialized
33604 (estimate_ipcp_clone_size_and_time): Update.
33605 (inline_merge_summary): Update.
33606 (do_estimate_edge_time): Update.
33607 (do_estimate_edge_size): Update.
33608 (do_estimate_edge_hints): Update.
33609 (inline_read_section, inline_write_summary): Stream both new predicates.
33610 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
33612 (compute_inlined_call_time): Cleanup.
33613 (big_speedup_p): Update.
33614 (edge_badness): Update.
33615 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
33616 (size_time_entry): Replace predicate by exec_predicate and
33617 nonconst_predicate.
33618 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
33619 (estimate_edge_time): Return also nonspec_time.
33620 (reset_edge_growth_cache): Update.
33622 2017-04-29 Jakub Jelinek <jakub@redhat.com>
33624 PR rtl-optimization/80491
33625 * ifcvt.c (noce_process_if_block): When looking for x setter
33626 with missing else_bb, don't check only the insn right before
33627 cond_earliest, but look for the last insn that x is modified in
33628 within the same bb.
33630 PR rtl-optimization/80491
33631 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
33633 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
33635 PR tree-optimization/80487
33636 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
33638 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
33640 PR tree-optimization/79697
33641 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
33642 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
33643 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
33645 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
33646 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
33648 2017-04-28 Martin Sebor <msebor@redhat.com>
33650 PR tree-optimization/80523
33651 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
33652 (init_target_to_host_charmap, target_to_host, target_strtol10): New
33654 (maybe_warn, format_directive, parse_directive): Use new functions.
33655 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
33657 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
33659 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
33661 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
33663 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
33664 target_header_dir): Set correctly.
33665 * configure: Regenerated.
33666 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
33667 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
33668 instead of SYSTEM_HEADER_DIR.
33670 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
33672 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
33673 (estimate_local_effects): Likewise.
33674 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
33675 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
33676 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
33677 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
33678 do_estimate_edge_time, estimate_edge_time): Likewise.
33679 * ipa-inline-analysis.c (estimate_node_size_and_time,
33680 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
33681 (estimate_time_after_inlining): Remove.
33683 2017-04-28 Martin Liska <mliska@suse.cz>
33685 * doc/gcov.texi: Enhance documentation of gcov.
33687 2017-04-28 Martin Liska <mliska@suse.cz>
33689 * doc/gcov.texi: Sort options in alphabetic order.
33690 * doc/gcov-dump.texi: Likewise.
33691 * doc/gcov-tool.texi: Likewise.
33692 * gcov.c (print_usage): Likewise.
33693 * gcov-dump.c (print_usage): Likewise.
33694 * gcov-tool.c (print_merge_usage_message): Likewise.
33695 (print_rewrite_usage_message): Likewise.
33696 (print_overlap_usage_message): Likewise.
33698 2017-04-28 Martin Liska <mliska@suse.cz>
33700 PR gcov-profile/53915
33701 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
33703 2017-04-28 Martin Liska <mliska@suse.cz>
33705 PR gcov-profile/79891
33706 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
33707 is marked by compiler as living on a line.
33708 (get_cycles_count): Remove usage of the union.
33709 (output_intermediate_file): Likewise.
33710 (find_source): Fix GNU coding style.
33711 (accumulate_line_counts): Remove old non-all block mode.
33712 (output_lines): Remove usage of the union.
33713 * profile.c (output_location): Include all BBs, even if
33714 belonging to a same line (and file) as a previous BB.
33716 2017-04-28 Martin Liska <mliska@suse.cz>
33718 * gcov.c (process_args): Handle new argument 'w'.
33719 (read_graph_file): Assign ID to BBs.
33720 (output_branch_count): Display BB # if verbose flag is set.
33721 (output_lines): Likewise for arcs.
33722 (print_usage): Add '--verbose' option help.
33723 * doc/gcov.texi: Document --verbose (-w) option.
33725 2017-04-28 Martin Liska <mliska@suse.cz>
33727 * gcov.c (struct block_location_info): New struct.
33728 (process_file): Fill up the new structure.
33729 (read_graph_file): Replace usage of encoding by the newly added
33731 (add_line_counts): Likewise.
33732 (accumulate_line_counts): Remove usage of the union.
33733 (function_info::function_info): New function.
33734 (function_info::~function_info): Likewise.
33735 (process_file): Call delete instead of release_function.
33736 (release_function): Release the function.
33737 (release_structures): Call delete instead of release_function.
33738 (solve_flow_graph): Replace usage of num_blocks.
33739 (find_exception_blocks): Likewise.
33740 (output_lines): Fix GNU coding style.
33742 2017-04-28 Martin Liska <mliska@suse.cz>
33745 * coverage.c (coverage_remove_note_file): New function.
33746 * coverage.h: Declare the function.
33747 * toplev.c (finalize): Clean if an error has been seen.
33749 2017-04-28 Martin Liska <mliska@suse.cz>
33751 PR gcov-profile/80031
33752 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
33753 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
33754 * gcov.c (read_graph_file): Read just number of blocks.
33755 * profile.c (branch_prob): Do not stream 0 flags per a basic
33758 2017-04-28 Martin Liska <mliska@suse.cz>
33760 * gcov-dump.c (tag_*): Add new argument to declarations.
33761 (dump_gcov_file): Likewise.
33762 (tag_blocks): Add and use new argument depth.
33763 (tag_arcs): Likewise.
33764 (tag_lines): Likewise.
33765 (tag_counters): Likewise.
33766 (tag_summary): Likewise.
33767 (dump_working_sets): Use depth to do a proper indentation.
33769 2017-04-28 Jakub Jelinek <jakub@redhat.com>
33772 * cgraph.h (symtab_node::debug_symtab): No longer inline.
33773 * symtab.c (symtab_node::debug_symtab): Move definition here.
33775 2017-04-28 Richard Biener <rguenther@suse.de>
33777 * lto-streamer.h (LTO_major_version): Bump to 7.
33779 2017-04-28 Richard Biener <rguenther@suse.de>
33781 * tree-vrp.c (assert_info): New struct.
33782 (add_assert_info): New helper.
33783 (register_edge_assert_for_2): Refactor to add asserts to a vector
33785 (register_edge_assert_for_1): Likewise.
33786 (register_edge_assert_for): Likewise.
33787 (finish_register_edge_assert_for): New helper actually registering
33788 asserts where live on edge.
33789 (find_conditional_asserts): Adjust.
33790 (find_switch_asserts): Likewise.
33791 (evrp_dom_walker::try_find_new_range): Generalize.
33792 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
33794 2017-04-27 Marek Polacek <polacek@redhat.com>
33797 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
33798 arg10 and arg11 to itype.
33800 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
33802 * doc/extend.texi (Object Size Checking): Improve grammar.
33804 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
33807 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
33808 that the logic for permitting reciprocal estimates matches that
33811 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33814 * tree.c (type_cache_hasher::equal): Only compare
33815 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
33816 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
33817 non-aggregate element types.
33818 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
33819 about the flag on ARRAY_TYPEs in the comment, formatting fix.
33821 2017-04-27 Richard Biener <rguenther@suse.de>
33823 PR middle-end/80533
33824 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
33825 stripping ARRAY_REFs from MEM_EXPR make sure we're not
33826 keeping a reference to a trailing array.
33828 2017-04-27 Richard Biener <rguenther@suse.de>
33830 PR middle-end/80539
33831 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
33832 being in loop-closed SSA form conservatively.
33833 (chrec_fold_multiply_poly_poly): Likewise.
33835 2017-04-27 Tamar Christina <tamar.christina@arm.com>
33837 PR middle-end/79665
33838 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
33839 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
33841 2017-04-27 Jakub Jelinek <jakub@redhat.com>
33844 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
33845 (aarch64_function_arg_alignment): Return unsigned int again, but still
33846 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
33847 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33848 Don't emit -Wpsabi note.
33849 (aarch64_function_arg_boundary): Likewise.
33850 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33853 2017-04-26 Nathan Sidwell <nathan@acm.org>
33855 * tree.h (crc32_unsigned_n): Declare.
33856 (crc32_unsigned, crc32_unsigned): Make inline.
33857 * tree.c (crc32_unsigned_bits): Replace with ...
33858 (crc32_unsigned_n): ... this.
33859 (crc32_unsigned, crc32_byte): Remove.
33860 (crc32_string): Remove unnecessary braces.
33862 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33864 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
33865 * ipa-inline-analysis.c (MAX_TIME): Remove.
33866 (account_size_time): Use sreal for time.
33867 (dump_inline_summary): Update.
33868 (estimate_function_body_sizes): Update.
33869 (estimate_edge_size_and_time): Update.
33870 (estimate_calls_size_and_time): Update.
33871 (estimate_node_size_and_time): Update.
33872 (inline_merge_summary): Update.
33873 (inline_update_overall_summary): Update.
33874 (estimate_time_after_inlining): Update.
33875 (inline_read_section): Update.
33876 (inline_write_summary): Update.
33877 * ipa-inline.c (compute_uninlined_call_time): Update.
33878 (compute_inlined_call_time): Update.
33879 (recursive_inlining): Update.
33880 (inline_small_functions): Update.
33881 (dump_overall_stats): Update.
33882 * ipa-inline.h: Include sreal.h.
33883 (size_time_entry): Turn time to sreal.
33884 (inline_summary): Turn self_time nad time to sreal.
33886 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
33888 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
33890 (sreal::stream_out, sreal::stream_in): New.
33891 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
33893 2017-04-25 Jakub Jelinek <jakub@redhat.com>
33895 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
33898 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
33901 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
33903 (dimode_scalar_chain::compute_convert_gain): Ditto.
33904 (dimode_scalar_chain::make_vector_copies): Ditto.
33905 (dimode_scalar_chain::convert_reg): Ditto.
33906 (dimode_scalar_chain::convert_insn): Ditto.
33907 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
33908 (VI248_AVX512BW_1): New mode iterator.
33909 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
33910 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
33913 2017-04-25 Martin Sebor <msebor@redhat.com>
33915 PR tree-optimization/80497
33916 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
33917 constants are representable in HOST_WIDE_INT.
33918 (parse_directive): Ditto.
33920 2017-04-25 Martin Sebor <msebor@redhat.com>
33923 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
33924 (new_zero_array): Adjust signature.
33925 (dom_info::dom_init): Used unsigned rather that size_t.
33926 (dom_info::dom_info): Same.
33928 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33929 Jakub Jelinek <jakub@redhat.com>
33932 * config/arm/arm.c: Include gimple.h.
33933 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33934 returns negative, increment ncrn only if it returned positive.
33935 (arm_needs_doubleword_align): Return int instead of bool,
33936 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
33937 members, but if there is any such non-FIELD_DECL
33938 > PARM_BOUNDARY aligned decl, return -1 instead of false.
33939 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
33940 returns negative, increment nregs only if it returned positive.
33941 (arm_setup_incoming_varargs): Likewise.
33942 (arm_function_arg_boundary): Emit -Wpsabi note if
33943 arm_needs_doubleword_align returns negative, return
33944 DOUBLEWORD_ALIGNMENT only if it returned positive.
33946 2017-04-25 Marek Polacek <polacek@redhat.com>
33949 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
33950 first argument to type.
33952 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
33955 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
33956 type checks to test for compatibility instead of equality.
33958 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
33959 Jakub Jelinek <jakub@redhat.com>
33962 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
33964 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
33965 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
33966 the alignment computation, but return their maximum in warn_alignment.
33967 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
33968 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
33970 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
33971 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
33974 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33976 * config/arc/simdext.md (dmpyh): Fix typo.
33978 2017-04-25 Richard Biener <rguenther@suse.de>
33980 PR tree-optimization/80492
33981 * alias.c (compare_base_decls): Handle registers with asm
33982 specification conservatively.
33983 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
33984 compare_base_decls returning dont-know properly.
33986 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33988 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
33989 (legitimate_offset_address_p): New function.
33990 (arc_legitimate_address_p): Use above function.
33992 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33994 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
33996 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
33998 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
33999 ACCH registers whenever they are available.
34001 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34003 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
34004 double regs fix when not used.
34006 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34008 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
34010 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
34011 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
34013 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34015 * config/arc/arc.c (arc_output_addsi): Check for h-register class
34016 when emitting short ADD instructions.
34018 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
34020 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
34022 (cmpsi_cc_c_insn): Likewise.
34023 (cbranchsi4_scratch): Compute proper instruction length using
34024 compact_hreg_operand.
34025 * config/arc/predicates.md (compact_hreg_operand): New predicate.
34027 2017-04-25 Richard Biener <rguenther@suse.de>
34029 PR middle-end/80509
34030 * passes.c (pass_manager::pass_manager): Initialize
34031 m_name_to_pass_map.
34033 2017-04-25 Richard Biener <rguenther@suse.de>
34035 PR tree-optimization/79201
34036 * tree-ssa-sink.c (statement_sink_location): Handle calls.
34038 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34041 * config/s390/vector.md: Split MEM->GPR vector moves for
34042 non-s_operand addresses.
34044 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
34047 * config/s390/predicates.md (reload_const_wide_int_operand): New
34049 * config/s390/s390.md ("movti"): Remove d/P alternative.
34050 ("movti_bigconst"): New pattern definition.
34052 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34055 * s390-protos.h (s390_expand_cs_hqi): Removed.
34056 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
34057 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
34058 modes as well as CCZ1mode and CCZmode.
34059 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
34060 signature of s390_emit_compare_and_swap.
34061 (s390_expand_cs_hqi): Likewise, make static.
34062 (s390_expand_cs_tdsi): Generate an explicit compare before trying
34063 compare-and-swap, in some cases.
34064 (s390_expand_cs): Wrapper function.
34065 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
34067 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
34068 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
34069 patterns for small and large integers. Forbid symref memory operands.
34070 Move expander to s390.c. Require cc register.
34071 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
34072 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
34073 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
34074 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
34075 symref memory operands. Remove CC mode and call s390_match_ccmode
34077 ("atomic_exchange<mode>"): Allow and implement all integer modes.
34079 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34081 * config/s390/s390.md (define_peephole2): New peephole to help
34082 combining the load-and-test pattern with volatile memory.
34084 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
34086 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
34087 with CCZmode for TARGET_Z196.
34089 2017-04-25 Jakub Jelinek <jakub@redhat.com>
34091 PR rtl-optimization/80501
34092 * combine.c (make_compound_operation_int): Set subreg_code to SET
34093 even for AND with mask of the sign bit of mode.
34095 PR rtl-optimization/80500
34096 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
34097 sum's initial value.
34099 2017-04-25 Julian Brown <julian@codesourcery.com>
34100 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34102 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
34104 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
34106 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
34108 2017-04-25 Julian Brown <julian@codesourcery.com>
34109 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34111 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
34112 (thunderx2t99_sha): New Reservation.
34114 2017-04-25 Julian Brown <julian@codesourcery.com>
34115 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
34117 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
34118 type for 1-element load.
34120 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
34122 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
34124 2017-04-24 Martin Jambor <mjambor@suse.cz>
34126 PR tree-optimization/80293
34127 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
34128 char arrays not totally scalarizable if it is false.
34129 (analyze_all_variable_accesses): Pass correct value in the new
34130 parameter. Add a statistics counter.
34132 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
34134 PR middle-end/79931
34135 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
34137 2017-04-24 Richard Biener <rguenther@suse.de>
34139 PR tree-optimization/80494
34140 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
34141 out for complex types.
34143 2017-04-24 Richard Biener <rguenther@suse.de>
34145 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
34146 * tree-ssa-sccvn.c (print_scc): Print SCC size.
34147 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
34148 (DFS): Adjust and never fail.
34149 (sccvn_dom_walker::fail): Remove.
34150 (sccvn_dom_walker::before_dom_children): Adjust.
34151 (run_scc_vn): Likewise and never fail.
34152 * tree-ssa-pre.c (pass_pre::execute): Adjust.
34153 (pass_fre::execute): Likewise.
34155 2017-04-24 Richard Biener <rguenther@suse.de>
34157 PR tree-optimization/79725
34158 * tree-ssa-sink.c (statement_sink_location): Return whether
34159 failure reason was zero uses. Move that check later.
34160 (sink_code_in_bb): Deal with zero uses by removing the stmt
34163 2017-04-24 Richard Biener <rguenther@suse.de>
34166 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
34167 pointer-based references.
34169 2017-04-24 Richard Biener <rguenther@suse.de>
34172 * pass_manager.h (pass_manager::operator new): Remove.
34173 (pass_manager::operator delete): Likewise.
34174 * passes.c (pass_manager::operator new): Remove.
34175 (pass_manager::operator delete): Likewise.
34176 (pass_manager::pass_manager): Zero individual pass members.
34178 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
34181 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
34182 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
34183 Check "XEXP (src, 1)" operand here.
34184 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
34185 Check "XEXP (src, 1)" operand here.
34186 (dimode_scalar_chain::make_vector_copies): Detect count register
34187 of a shift instruction. Zero extend count register from QImode
34188 to DImode to satisfy vector shift pattern count operand predicate.
34189 Substitute vector shift count operand with a DImode copy.
34190 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
34193 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
34195 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
34196 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34197 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
34198 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
34199 (UNSPEC_NOREX_MEM): Remove definition.
34201 2017-04-21 Richard Biener <rguenther@suse.de>
34203 PR tree-optimization/79547
34204 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
34205 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
34206 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
34207 without any constraints.
34209 2017-04-21 Richard Biener <rguenther@suse.de>
34211 PR tree-optimization/78847
34212 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
34214 2017-04-21 Richard Biener <rguenther@suse.de>
34216 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
34217 (build_distinct_type_copy): Likewise.
34218 (build_variant_type_copy): Likewise.
34219 * tree.c (build_qualified_type): Pass down mem-stat info.
34220 (build_distinct_type_copy): Likewise.
34221 (build_variant_type_copy): Likewise.
34223 2017-04-21 Richard Biener <rguenther@suse.de>
34225 PR tree-optimization/80237
34226 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
34228 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
34229 for a simplified result.
34231 2016-04-21 Richard Biener <rguenther@suse.de>
34233 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
34234 sth as strict as a simple_iv but a chrec without symbols and an
34235 operand defined in the loop we are peeling (and not some subloop).
34236 (propagate_constants_for_unrolling): Propagate all constants.
34238 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34241 * config/i386/i386.c (print_reg): Remove assert for disalowed
34242 regno values, call output_operand_lossage instead.
34244 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
34247 * config/i386/constraints.md (Yc): New register constraint.
34248 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
34249 Use Yc constraint for alternative 2 of operand 0. Remove
34250 preferred_for_speed attribute.
34252 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34254 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
34255 lastprivate clauses in SIMT case.
34257 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
34259 * doc/invoke.texi (-Wextra-semi): Document new warning option.
34261 2017-04-20 Richard Biener <rguenther@suse.de>
34263 PR tree-optimization/57796
34264 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
34265 as N scalar stores.
34266 (vect_model_load_cost): Cost gathers as N scalar loads.
34268 2017-04-20 Richard Biener <rguenther@suse.de>
34270 * ggc-page.c (ggc_allocated_p): Rename to ...
34271 (safe_lookup_page_table_entry): ... this and return the lookup
34273 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
34275 2017-04-20 Richard Biener <rguenther@suse.de>
34277 PR tree-optimization/80453
34278 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
34279 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
34280 from the conditions.
34281 (vn_phi_eq): Pass them down.
34282 (vn_phi_lookup): Record them.
34283 (vn_phi_insert): Likewise.
34285 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
34287 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
34288 uninitialized variable warning to avoid buffer overrun.
34290 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
34293 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
34294 is suppressed for '{ 0 }' in C.
34296 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34298 * BASE-VER: Set to 8.0.0.
34300 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
34302 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
34303 priority .init_array and .fini_array section with SECTION_NOTYPE
34306 2017-04-20 Jakub Jelinek <jakub@redhat.com>
34308 PR middle-end/80423
34309 * tree.h (build_array_type): Add typeless_storage default argument.
34310 * tree.c (type_cache_hasher::equal): Also compare
34311 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
34312 (build_array_type): Add typeless_storage argument, set
34313 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
34315 (build_nonshared_array_type): Adjust build_array_type_1 caller.
34316 (build_array_type): Likewise. Add typeless_storage argument.
34318 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34319 Jakub Jelinek <jakub@redhat.com>
34321 PR tree-optimization/80426
34322 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
34323 operation on symbolic operands, also compute the overflow for the
34324 invariant part when the operation degenerates into a negation.
34326 2017-04-19 Jakub Jelinek <jakub@redhat.com>
34329 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
34330 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
34333 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
34335 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
34338 * config/avr/avr.c (tree.h): Include it.
34339 (cgraph.h): Include it.
34340 (avr_encode_section_info): Don't warn for uninitialized progmem
34341 variable if it's just an alias.
34343 2017-04-19 Richard Biener <rguenther@suse.de>
34346 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
34347 when needed by AutoPGO.
34349 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
34352 * doc/lto.texi: Remove an extra 'that'.
34354 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
34356 PR rtl-optimization/80429
34357 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
34358 are only used in debug insns.
34360 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
34361 Vladimir Makarov <vmakarov@redhat.com>
34363 * config/sparc/predicates.md (input_operand): Add comment. Return
34364 true for any memory operand when LRA is in progress.
34365 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
34367 2017-04-18 Jeff Law <law@redhat.com>
34370 * mips.md ({return,simple_return}_internal): Do not overwrite
34373 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34375 PR tree-optimization/80443
34376 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
34377 instead of adding 1, subtract -1 and similarly instead of subtracting
34380 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
34382 PR rtl-optimization/80357
34383 * haifa-sched.c (tmp_bitmap): New variable.
34384 (model_recompute): Handle duplicate use records.
34385 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
34386 (free_global_sched_pressure_data): Free it.
34388 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
34391 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
34392 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
34393 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
34394 instead of SYSTEM_HEADER_DIR.
34396 2017-04-18 Jeff Law <law@redhat.com>
34398 PR middle-end/80422
34399 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
34400 predecessors after walking up the insn chain.
34402 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34405 * dwarf2out.c (modified_type_die): Try harder not to emit internal
34406 sizetype type into debug info.
34408 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
34411 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
34412 unneeded test for TARGET_UPPER_REGS_SF.
34413 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
34415 2017-04-18 Jakub Jelinek <jakub@redhat.com>
34418 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
34419 instead of gsi_after_labels.
34421 2017-04-18 Jeff Law <law@redhat.com>
34423 * regcprop.c (maybe_mode_change): Avoid creating copies of the
34427 2017-04-13 Jeff Law <law@redhat.com>
34428 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34429 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34431 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
34434 * config/avr/avr.c (intl.h): Include it.
34435 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
34437 2017-04-18 Martin Liska <mliska@suse.cz>
34439 PR gcov-profile/78783
34440 * gcov-tool.c (gcov_output_files): Validate that destination
34441 file is either removed by the tool or by a user.
34443 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34444 Guy Benyei <guybe@mellanox.com>
34446 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
34447 block, and do not negate it, the stored id is already negative.
34449 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
34451 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
34453 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
34456 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
34457 masks of options that should be turned off if the VSX vector
34458 options are turned off.
34459 (OTHER_P8_VECTOR_MASKS): Likewise.
34460 (OTHER_VSX_VECTOR_MASKS): Likewise.
34461 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
34462 rs6000_disable_incompatible_switches to validate no type switches
34464 (rs6000_incompatible_switch): New function to disallow turning on
34465 other vector options if -mno-vsx, -mno-power8-vector, or
34466 -mno-power9-vector are specified.
34468 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34470 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
34472 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34474 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
34475 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
34476 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
34477 (ARG_POINTER_CFA_OFFSET): Likewise.
34479 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
34481 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
34482 conditions to take advantage of various optimizations.
34484 2017-04-13 Jeff Law <law@redhat.com>
34486 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
34487 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
34488 (zero_extendsidi2_dext): Likewise.
34490 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34493 * fold-const.c (fold_ternary_loc): Revert
34494 use op0 instead of fold_convert_loc (loc, type, arg0) part of
34497 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
34499 PR rtl-optimization/80343
34500 * lra-remat.c (update_scratch_ops): Assign original hard reg to
34501 new scratch pseudo.
34503 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
34506 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
34507 to ubsan_encode_value.
34509 2017-04-13 Jeff Law <law@redhat.com>
34511 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
34512 appearing in DEBUG_INSNs.
34514 2017-04-13 Martin Liska <mliska@suse.cz>
34516 PR gcov-profile/80413
34517 * gcov-io.c (gcov_write_string): Copy to buffer just when
34518 allocated size is greater than zero.
34520 2017-04-13 Jakub Jelinek <jakub@redhat.com>
34523 * dwarf2out.c (decls_for_scope): Ignore declarations of
34524 current_function_decl in BLOCK_NONLOCALIZED_VARS.
34526 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
34529 * ipa-visibility.c (non_local_p): Fix typos.
34530 (localize_node): When localizing symbol in same comdat group,
34531 dissolve the group only when we know external symbols are going
34533 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
34535 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34537 PR tree-optimization/79390
34538 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
34539 order does not result in usable sequence, retry with reversed operand
34545 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
34546 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
34547 op0 instead of fold_convert_loc (loc, type, arg0).
34549 2017-04-12 Jeff Law <law@redhat.com>
34551 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
34552 has a delay slot in the generated code.
34554 * config/cris/cris.md (cris_preferred_reload_class): Return
34555 GENNONACR_REGS rather than GENERAL_REGS.
34557 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34560 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
34561 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
34562 signedness of the result type.
34564 2017-04-12 Richard Biener <rguenther@suse.de>
34565 Jeff Law <law@redhat.com>
34567 PR tree-optimization/80359
34568 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
34569 trim stores to TARGET_MEM_REFs.
34571 2017-04-12 Richard Biener <rguenther@suse.de>
34573 PR tree-optimization/79390
34574 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
34575 threading case even more.
34577 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
34580 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
34581 for quad_address_p for TImode, instead of just not indexed_address.
34583 2017-04-12 Richard Biener <rguenther@suse.de>
34584 Bernd Edlinger <bernd.edlinger@hotmail.de>
34586 PR middle-end/79671
34587 * alias.c (component_uses_parent_alias_set_from): Handle
34588 TYPE_TYPELESS_STORAGE.
34589 (get_alias_set): Likewise.
34590 * tree-core.h (tree_type_common): Add typeless_storage flag.
34591 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
34592 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
34593 for types containing members with TYPE_TYPELESS_STORAGE.
34594 (place_field): Likewise.
34595 (layout_type): Likewise for ARRAY_TYPE.
34596 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
34597 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
34598 TYPE_TYPELESS_STORAGE.
34599 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
34601 2017-04-12 Jakub Jelinek <jakub@redhat.com>
34604 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
34605 first argument to type.
34607 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34611 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
34612 CONST0_RTX (mode) rather than const0_rtx where appropriate.
34613 (rs6000_expand_binop_builtin): Likewise.
34614 (rs6000_expand_ternop_builtin): Likewise; also add missing
34615 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
34616 vshasigma built-ins.
34617 * doc/extend.texi: Document that vec_xxpermdi's third argument
34618 must be a constant.
34620 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
34622 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
34623 Use shift_const cost parameter when calculating gain of STV shifts.
34625 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
34627 PR rtl-optimization/70478
34628 * lra-constraints.c (process_alt_operands): Check memory for
34629 disfavoring memory insn operand.
34631 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34633 PR middle-end/80100
34634 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
34635 left shift in unsigned HOST_WIDE_INT type.
34637 PR rtl-optimization/80385
34638 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
34639 (not (neg X)) into (plus X -1) for complex or non-integral modes.
34642 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
34643 if they have any depend clauses.
34645 2017-04-11 Martin Liska <mliska@suse.cz>
34648 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
34649 * ipa-split.c (split_function): Create a local comdat symbol
34650 if caller is in a comdat group.
34652 2017-04-11 Martin Liska <mliska@suse.cz>
34655 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
34658 2017-04-11 Martin Sebor <msebor@redhat.com>
34660 PR middle-end/80364
34661 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
34662 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
34664 (directive::set_width, directive::set_precision, format_character):
34666 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
34669 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
34672 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
34673 conflict, set target->arch_name instead of target->cpu_name.
34675 2017-04-11 Richard Biener <rguenther@suse.de>
34677 PR tree-optimization/80374
34678 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
34679 build_zero_cst, remove fold_convertible_p check again.
34681 2017-04-11 Martin Liska <mliska@suse.cz>
34684 * ubsan.c (instrument_object_size): Do not instrument register
34687 2017-04-11 Jakub Jelinek <jakub@redhat.com>
34690 * config/i386/i386-builtin-types.def
34691 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
34692 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
34693 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
34694 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
34695 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
34696 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
34697 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
34698 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
34699 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
34700 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
34701 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
34702 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
34703 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
34704 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
34705 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
34706 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
34707 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
34708 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
34709 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
34710 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
34711 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
34712 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
34713 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
34714 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
34715 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
34716 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
34717 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
34718 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
34719 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
34720 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
34721 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
34722 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
34723 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
34724 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
34725 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
34726 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
34727 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
34728 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
34729 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
34730 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
34731 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
34732 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
34733 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
34734 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
34735 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
34736 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
34738 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
34739 flag to second_arg_count, handle 4 argument function type _COUNT
34740 aliases, handle second_arg_count on second argument rather than last.
34742 2017-04-10 Jeff Law <law@redhat.com>
34744 PR tree-optimization/80374
34745 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
34746 record anything if we can not convert integer_zero_node to the
34749 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
34752 * config/rs6000/rs6000.c (rs6000_option_override_internal):
34753 Enhance special handling given to the TARGET_P9_MINMAX option in
34754 relation to certain other options.
34756 2017-04-10 Bin Cheng <bin.cheng@arm.com>
34758 PR tree-optimization/80153
34759 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
34760 remove POINTER_PLUS_EXPR's base part directly, rather than through
34763 2017-04-10 Richard Biener <rguenther@suse.de>
34764 Bin Cheng <bin.cheng@arm.com>
34766 PR tree-optimization/80153
34767 * tree-affine.c (aff_combination_to_tree): Get base pointer from
34768 the first element of pointer type aff_tree. Build result expr in
34770 (add_elt_to_tree): Convert to type unconditionally. Remove other
34771 fold_convert calls.
34772 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
34773 (rewrite_use_nonlinear_expr): Check invariant using iv information.
34775 2017-04-10 Richard Biener <rguenther@suse.de>
34777 * tree-ssa-structalias.c (find_func_aliases): Properly handle
34780 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
34782 PR rtl-optimization/70478
34783 * lra-constraints.c (curr_small_class_check): New.
34784 (update_and_check_small_class_inputs): New.
34785 (process_alt_operands): Update curr_small_class_check. Disfavor
34786 alternative insn memory operands. Check available regs for small
34789 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
34792 * config/mips/mips.opt (-mvirt): Update description.
34793 * doc/invoke.texi (-mvirt): Likewise.
34795 2017-04-10 Richard Biener <rguenther@suse.de>
34797 PR middle-end/80362
34798 * fold-const.c (fold_binary_loc): Look at unstripped ops when
34799 looking for NEGATE_EXPR in -A / -B to A / B folding.
34801 2017-04-10 Martin Liska <mliska@suse.cz>
34803 PR gcov-profile/80224
34804 * gcov.c (print_usage): Fix usage string.
34805 (get_gcov_intermediate_filename): Remove.
34806 (output_gcov_file): Use both for normal and intermediate format.
34807 (generate_results): Do not initialize special file for
34808 intermediate format.
34810 2017-04-10 Richard Biener <rguenther@suse.de>
34812 PR tree-optimization/80304
34813 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
34816 2017-04-10 Nathan Sidwell <nathan@acm.org>
34819 * config/rs6000/rs6000.c (rs6000_vector_type): New.
34820 (rs6000_init_builtins): Use it.
34822 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34824 * config/arm/arm.md (<mrc>): Add mode to SET source.
34825 (<mrrc>): Likewise.
34827 2017-04-10 Richard Biener <rguenther@suse.de>
34829 PR middle-end/80344
34830 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
34832 2017-04-10 Jakub Jelinek <jakub@redhat.com>
34835 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
34836 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
34837 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
34838 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
34839 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
34840 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
34841 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
34842 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
34843 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
34844 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
34845 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
34846 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
34847 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
34848 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
34849 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
34850 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
34851 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
34852 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
34853 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
34854 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
34855 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
34856 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
34857 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
34859 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
34861 PR rtl-optimization/70478
34862 * lra-constraints.c: Reverse the last patch.
34864 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
34866 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
34867 Add comment for WCHAR_T.
34869 2017-04-08 Martin Liska <mliska@suse.cz>
34872 2017-04-07 Martin Liska <mliska@suse.cz>
34875 * ipa-split.c (split_function): Add function part to a same comdat
34878 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
34881 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
34883 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
34885 * rs6000/rs6000.c (vec_load_pendulum): Rename...
34886 (vec_pairing): ...to this.
34887 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
34888 (rs6000_sched_init): Adjust for name change.
34889 (struct rs6000_sched_context): Likewise.
34890 (rs6000_init_sched_context): Likewise.
34891 (rs6000_set_sched_context): Likewise.
34893 2017-04-07 Jakub Jelinek <jakub@redhat.com>
34899 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
34901 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
34902 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
34903 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
34905 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
34907 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
34909 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34911 PR rtl-optimization/70703
34912 * ira-color.c (update_conflict_hard_regno_costs): Use
34913 int64_t instead of HOST_WIDE_INT.
34915 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
34917 PR rtl-optimization/70478
34918 * lra-constraints.c (process_alt_operands): Disfavor alternative
34919 insn memory operands.
34921 2017-04-07 Jeff Law <law@redhat.com>
34923 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
34924 CALL and NOTE_INSN_CALL_ARG_LOCATION.
34926 2017-04-07 Martin Liska <mliska@suse.cz>
34929 * config/aarch64/aarch64.c (aarch64_process_target_attr):
34930 Show error message instead of an ICE.
34932 2017-04-07 Martin Liska <mliska@suse.cz>
34935 * ipa-split.c (split_function): Add function part to a same comdat
34938 2017-04-07 Richard Biener <rguenther@suse.de>
34940 PR middle-end/80341
34941 * tree.c (get_unwidened): Also handle ! for_type case for
34943 * convert.c (do_narrow): Split out from ...
34944 (convert_to_integer_1): ... here. Do not pass final truncation
34945 type to get_unwidened for TRUNC_DIV_EXPR.
34947 2017-04-07 Richard Biener <rguenther@suse.de>
34949 * tree-affine.c (wide_int_ext_for_comb): Take type rather
34951 (aff_combination_const): Adjust.
34952 (aff_combination_scale): Likewise.
34953 (aff_combination_add_elt): Likewise.
34954 (aff_combination_add_cst): Likewise.
34955 (aff_combination_convert): Likewise.
34956 (add_elt_to_tree): Likewise. Remove unused argument.
34957 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
34959 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
34961 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
34963 * config/arm/arm.c (arm_default_short_enums): Use
34964 ARM_DEFAULT_SHORT_ENUMS.
34965 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
34967 2017-04-06 Jakub Jelinek <jakub@redhat.com>
34970 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
34971 members with redundant out-of-class redeclaration.
34973 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34976 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
34977 * config/i386/i386.md (*zero_extendsidi2):
34978 Add (?*x,*x) and (?*v,*v) alternatives.
34980 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
34983 * config/i386/i386.c (ix86_expand_builtin)
34984 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
34985 mode from insn data. Convert operands to insn operand mode.
34986 Copy operands that don't satisfy insn predicate to a register.
34988 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
34990 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
34993 2017-04-06 Richard Biener <rguenther@suse.de>
34995 PR tree-optimization/80334
34996 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
34997 preserve alignment of accesses.
34999 2017-04-06 Richard Biener <rguenther@suse.de>
35001 PR tree-optimization/80262
35002 * tree-sra.c (build_ref_for_offset): Preserve address-space
35004 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
35005 Drop useless address-space information on MEM_REF offsets.
35007 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
35009 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
35011 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
35013 PR rtl-optimization/70703
35014 * ira-color.c (update_conflict_hard_regno_costs): Use
35015 HOST_WIDE_INT instead of long.
35017 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
35020 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
35021 not defined for x86_64 target. Add -mmmx target option when __SSE2__
35023 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
35024 for x86_64 target. Handle -m3dnowa option.
35026 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
35028 PR rtl-optimization/70703
35029 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
35030 (update_conflict_hard_regno_costs): Use long instead of unsigned
35031 arithmetic for cost calculation.
35033 2017-04-05 Jakub Jelinek <jakub@redhat.com>
35034 Bernd Edlinger <bernd.edlinger@hotmail.de>
35037 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
35040 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
35043 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
35044 ptr_mode with Pmode throughout.
35045 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
35046 into probe_stack_range and use DImode.
35048 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
35051 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
35052 call_eh_return is true.
35054 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35056 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
35057 Initialize last_match_fntype_index.
35059 2017-04-05 Jakub Jelinek <jakub@redhat.com>
35062 * tree-nvr.c: Include internal-fn.h.
35063 (pass_return_slot::execute): Ignore internal calls without
35066 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35067 Richard Biener <rguenther@suse.de>
35070 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
35071 captures used multiple times, except for the last use.
35072 * generic-match-head.c: Include gimplify.h.
35074 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35076 PR tree-optimization/79390
35077 * target.h (struct noce_if_info): Declare.
35078 * targhooks.h (default_noce_conversion_profitable_p): Declare.
35079 * target.def (noce_conversion_profitable_p): New target hook.
35080 * ifcvt.h (struct noce_if_info): New type, moved from ...
35081 * ifcvt.c (struct noce_if_info): ... here.
35082 (noce_conversion_profitable_p): Renamed to ...
35083 (default_noce_conversion_profitable_p): ... this. No longer
35085 (noce_try_store_flag_constants, noce_try_addcc,
35086 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
35087 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
35088 instead of noce_conversion_profitable_p.
35089 * config/i386/i386.c: Include ifcvt.h.
35090 (ix86_option_override_internal): Don't override
35091 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
35092 (ix86_noce_conversion_profitable_p): New function.
35093 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
35094 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
35095 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
35096 * doc/tm.texi: Regenerated.
35098 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35100 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
35103 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
35106 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
35107 instructions for small multiply cores.
35109 2017-04-04 Jeff Law <law@redhat.com>
35111 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
35113 (mips_expand_vec_perm_const): Initialize elements in orig_perm
35114 that are not set by the loop over the elements.
35116 2017-04-04 Jakub Jelinek <jakub@redhat.com>
35119 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
35120 int mode, convert_modes it to mode as unsigned, otherwise use
35121 lowpart_subreg to mode rather than SImode.
35122 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
35123 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
35124 Use DImode instead of SImode for the shift count operand.
35125 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
35128 2017-04-04 Richard Biener <rguenther@suse.de>
35130 PR middle-end/80281
35131 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
35132 arithmetic done for the negate or the plus. Simplify.
35133 (A - (-B) -> A + B): Likewise.
35134 * fold-const.c (split_tree): Make sure to not negate pointers.
35136 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
35138 PR rtl-optimization/60818
35139 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
35140 a compare of comparisons with the thing compared if this results
35141 in a different machine mode.
35143 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
35145 * alias.c (base_alias_check): Fix typo in comment.
35146 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
35147 * cgraphunit.c (symbol_table::compile): Likewise.
35148 * collect2.c (maybe_run_lto_and_relink): Likewise.
35149 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
35150 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
35151 * config/avr/avr.c (avr_map_op_t): Likewise.
35152 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
35153 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
35154 * config/epiphany/epiphany.md (movcc): Likewise.
35155 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
35156 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
35158 * config/mips/mips.c (mips_save_restore_reg): Likewise.
35159 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
35160 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
35161 * config/sh/sh.c (sh_rtx_costs): Likewise.
35162 * fold-const.c (fold_truth_andor): Likewise.
35163 * genautomata.c (collapse_flag): Likewise.
35164 * gengtype.h (struct type::u::s): Likewise.
35165 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
35166 * input.c (FORMAT_AMOUNT): Likewise.
35167 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
35168 (known_aggs_to_agg_replacement_list): Likewise.
35169 * ipa-inline-analysis.c: Likewise.
35170 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
35171 * ipa-polymorphic-call.c
35172 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
35173 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
35174 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
35176 * modulo-sched.c (apply_reg_moves): Likewise.
35177 * omp-expand.c (build_omp_regions_1): Likewise.
35178 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
35179 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
35180 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
35181 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
35182 * value-prof.c: Likewise.
35183 * var-tracking.c (val_reset): Likewise.
35185 2017-04-03 Richard Biener <rguenther@suse.de>
35187 PR tree-optimization/80275
35188 * fold-const.c (split_address_to_core_and_offset): Handle
35191 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
35193 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
35194 descriptors is at least equal to that of functions.
35196 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35198 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
35200 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
35203 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
35204 (mov<IMOD4:mode>): New expander.
35205 (*mov<IMOD4:mode>_internal): New insn and split pattern.
35207 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
35209 PR rtl-optimization/79405
35210 * fwprop.c (propagations_left): New variable.
35211 (forward_propagate_into): Decrement it.
35212 (fwprop_init): Initialize it.
35213 (fw_prop): If the variable has reached zero, stop propagating.
35214 (fwprop_addr): Ditto.
35216 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35219 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
35220 a FUNCTION_DECL, pass it as decl instead of origin to
35223 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
35225 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
35228 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
35231 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
35232 TARGET_VSX_SMALL_INTEGER.
35234 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35236 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
35237 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
35239 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
35241 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
35242 extraction from odd-numbered MSA register.
35244 2017-03-31 Jakub Jelinek <jakub@redhat.com>
35246 PR middle-end/80173
35247 * expmed.c (store_bit_field_1): Don't attempt to create
35248 a word subreg out of hard registers wider than word if they
35249 have HARD_REGNO_NREGS of 1 for their mode.
35251 PR middle-end/80163
35252 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
35253 conversions to integer types wider than word and pointer.
35256 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
35257 (rtx_equal_for_cselib_p): Pass 0 to it.
35258 * cselib.c (cselib_hasher::equal): Likewise.
35259 (rtx_equal_for_cselib_1): Add depth argument. If depth
35260 is 128, don't look up VALUE locs and punt. Increment
35261 depth in recursive calls when walking VALUE locs.
35263 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
35265 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
35266 (make_gcov_file_name): Use the canonical path name for generating
35268 (read_line): Fix handling of files with ascii null bytes.
35270 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
35272 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
35273 to initialise a vector register instead
35274 of using a const_int.
35276 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35278 PR translation/80189
35279 * gimplify.c (omp_default_clause): Use %qs instead of %s in
35280 diagnostic messages.
35282 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
35285 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
35286 (dfp_diex_<mode>): Update mode of operand 1.
35287 * doc/extend.texi (dxex, dxexq): Document change to return type.
35288 (diex, diexq): Document change to argument type.
35290 2017-03-30 Martin Jambor <mjambor@suse.cz>
35293 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
35294 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
35295 it reflects the signature changes performed at the callee side.
35296 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
35297 to cgraph_build_function_type_skip_args.
35298 (build_function_decl_skip_args): Adjust call to the above function.
35300 2017-03-30 Jakub Jelinek <jakub@redhat.com>
35303 * config/i386/sse.md
35304 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
35305 register as dest whenever it is a MEM not rtx_equal_p to the
35306 corresponding dup operand, and when forcing into reg move the
35307 reg into the memory afterwards.
35308 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
35309 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
35310 for the force_reg mode.
35311 (avx512vl_vextractf128<mode>): Use register as dest either
35312 always when a MEM, or when it is a MEM not rtx_equal_p to the
35313 corresponding dup operand, or even not when it is a CONST_VECTOR
35314 depending on the mode and lo vs. hi.
35315 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
35317 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
35318 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
35319 Likewise. Require that operands[2] is even.
35320 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
35321 Remove extraneous parens. Require that operands[2] is a multiple
35323 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
35324 operands[0] is a MEM if <mask_applied>, the predicates/constraints
35325 disallow memory then.
35327 2017-03-30 Richard Biener <rguenther@suse.de>
35329 PR tree-optimization/77498
35330 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
35331 to non-constants over backedges.
35333 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
35335 PR rtl-optimization/80233
35336 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
35337 as last_combined_insn. Do not test for BARRIER_P separately.
35339 2017-03-29 Andreas Schwab <schwab@suse.de>
35342 * calls.c (prepare_call_address): Convert funexp to Pmode before
35343 copying to temp reg.
35345 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35347 PR tree-optimization/80158
35348 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
35349 Handle possible future case of more than one alternate
35351 (replace_rhs_if_not_dup): Likewise.
35352 (replace_one_candidate): Likewise.
35354 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
35356 PR rtl-optimization/80193
35357 * ira.c (ira): Do not check allocation for LRA.
35359 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
35361 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
35362 (nvptx_output_simt_exit): Declare.
35363 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
35364 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
35365 (init_softstack_frame): Move initialization of crtl->is_leaf to...
35366 (nvptx_declare_function_name): ...here. Emit declaration of local
35367 memory space buffer for omp_simt_enter insn.
35368 (nvptx_output_unisimt_switch): New.
35369 (nvptx_output_softstack_switch): New.
35370 (nvptx_output_simt_enter): New.
35371 (nvptx_output_simt_exit): New.
35372 * config/nvptx/nvptx.h (struct machine_function): New fields
35373 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
35374 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
35375 (UNSPECV_SIMT_EXIT): Ditto.
35376 (omp_simt_enter_insn): New insn.
35377 (omp_simt_enter): New expansion.
35378 (omp_simt_exit): New insn.
35379 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
35381 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
35382 (expand_GOMP_SIMT_ENTER_ALLOC): New.
35383 (expand_GOMP_SIMT_EXIT): New.
35384 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
35385 (GOMP_SIMT_ENTER_ALLOC): Ditto.
35386 (GOMP_SIMT_EXIT): Ditto.
35387 * target-insns.def (omp_simt_enter): New insn.
35388 (omp_simt_exit): Ditto.
35389 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
35391 (lower_rec_simd_input_clauses): Implement SIMT privatization.
35392 (lower_rec_input_clauses): Likewise.
35393 (lower_lastprivate_clauses): Handle SIMT privatization.
35395 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
35396 (ompdevlow_adjust_simt_enter): New.
35397 (find_simtpriv_var_op): New.
35398 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
35399 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
35401 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
35402 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
35403 (copy_decl_for_dup_finish): Ditto.
35405 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
35407 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
35410 * config/i386/i386.c (ix86_option_override_internal): Always
35411 allow -mpreferred-stack-boundary=3 for 64-bit targets.
35413 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35415 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
35417 2017-03-28 Bin Cheng <bin.cheng@arm.com>
35419 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
35420 mark new edge's irreducible flag accordign to it.
35421 (vect_do_peeling): Check loop preheader edge's irreducible flag
35422 and pass it to function slpeel_add_loop_guard.
35424 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
35426 PR tree-optimization/80218
35427 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
35428 Update block frequencies and counts.
35430 2017-03-28 Richard Biener <rguenther@suse.de>
35432 PR tree-optimization/78644
35433 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
35434 of a simplification result we may not use it at all.
35436 2017-03-28 Richard Biener <rguenther@suse.de>
35439 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
35440 without arguments, generate default definition of a SSA name.
35442 2017-03-28 Richard Biener <rguenther@suse.de>
35444 PR middle-end/80222
35445 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
35446 TYPE_REF_CAN_ALIAS_ALL references.
35447 * fold-const.c (fold_indirect_ref_1): Likewise.
35449 2017-03-28 Martin Liska <mliska@suse.cz>
35452 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
35453 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
35455 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35456 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
35458 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
35459 (EXTRA_SPECS): Define.
35460 (SUBTARGET_EXTRA_SPECS): Likewise.
35461 (SUBTARGET_CPP_SPEC): Likewise.
35462 * config/arc/elf.h (EXTRA_SPECS): Renamed to
35463 SUBTARGET_EXTRA_SPECS.
35464 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
35466 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
35468 * config/arc/simdext.md (vst64_insn): Update pattern.
35469 (vld32wh_insn): Likewise.
35470 (vld32wl_insn): Likewise.
35471 (vld64_insn): Likewise.
35472 (vld32_insn): Likewise.
35474 2017-03-28 Marek Polacek <polacek@redhat.com>
35477 * fold-const.c (fold_comparison): Use protected_set_expr_location
35478 instead of SET_EXPR_LOCATION.
35480 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
35482 * tree.c (add_expr): Avoid name lookup warning.
35484 2017-03-27 Jeff Law <law@redhat.com>
35486 PR tree-optimization/80216
35487 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
35488 function name. Limit recursion depth.
35489 (record_temporary_equivalences): Corresponding changes.
35491 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
35493 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
35496 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35499 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
35501 * cfgcleanup.c (reg_note_cfa_p): New array.
35502 (insns_have_identical_cfa_notes): New function.
35503 (old_insns_match_p): Don't cross-jump in between /f
35504 and non-/f instructions. If both i1 and i2 are frame related,
35505 verify all CFA notes, their order and content.
35507 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
35510 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
35511 HImode and SImode with zero extend to DImode to one insn.
35512 (bswap<mode>2_extenddi): Likewise.
35513 (bswapsi2_extenddi): Likewise.
35514 (bswaphi2_extendsi): Likewise.
35515 (bswaphi2): Combine bswap HImode and SImode into one insn.
35516 Separate memory insns from swapping register.
35517 (bswapsi2): Likewise.
35518 (bswap<mode>2): Likewise.
35519 (bswaphi2_internal): Delete, no longer used.
35520 (bswapsi2_internal): Likewise.
35521 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
35522 store, and gpr<-gpr swap insns.
35523 (bswap<mode>2_store): Likewise.
35524 (bswaphi2_reg): Register only splitter, combine with the splitter.
35525 (bswaphi2 splitter): Likewise.
35526 (bswapsi2_reg): Likewise.
35527 (bswapsi2 splitter): Likewise.
35528 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
35529 the insns into load, store, and register/register insns.
35530 (bswapdi2_ldbrx): Likewise.
35531 (bswapdi2_load): Likewise.
35532 (bswapdi2_store): Likewise.
35533 (bswapdi2_reg): Likewise.
35535 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
35537 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
35538 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
35540 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
35543 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
35545 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35546 special handling for target option conflicts between dform
35547 options (-mpower9-dform, -mpower9-dform-vector,
35548 -mpower9-dform-scalar) and -mno-direct-move.
35550 2017-03-27 Richard Biener <rguenther@suse.de>
35552 PR tree-optimization/80181
35553 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
35555 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35557 * config/arc/predicates.md (move_double_src_operand): Replace the
35558 call to move_double_src_operand with a call to address_operand.
35560 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35562 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
35563 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
35564 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
35566 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35568 * config/arc/predicates.md (long_immediate_loadstore_operand):
35569 Consider scaled addresses cases.
35571 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
35573 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
35574 restored when in interrupt.
35575 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
35576 doesn't have delay slot.
35578 2017-03-27 Richard Biener <rguenther@suse.de>
35581 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
35582 inlined thunk clones.
35584 2017-03-27 Jakub Jelinek <jakub@redhat.com>
35587 * asan.c (instrument_derefs): Copy over last operand from
35588 original COMPONENT_REF to the new COMPONENT_REF with
35589 DECL_BIT_FIELD_REPRESENTATIVE.
35590 * ubsan.c (instrument_object_size): Likewise.
35592 2017-03-27 Richard Biener <rguenther@suse.de>
35594 PR tree-optimization/80170
35595 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
35596 sure DR/SCEV didnt fold in constants we do not see when looking
35597 at the reference base alignment.
35599 2017-03-27 Richard Biener <rguenther@suse.de>
35601 PR middle-end/80171
35602 * gimple-fold.c (fold_ctor_reference): Properly guard against
35603 NULL return value from canonicalize_constructor_val.
35605 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
35608 * config/i386/i386.c (ix86_expand_builtin)
35609 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
35610 flags reg setting and flags reg using instructions.
35611 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
35612 clobbering instructions to zero extend op2.
35614 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
35616 * doc/install.texi (Configuration) <--with-aix-soname>:
35617 Update link to AIX ld.
35619 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
35621 PR rtl-optimization/80160
35622 PR rtl-optimization/80159
35623 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
35624 reg_alternate_class into account.
35626 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
35629 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
35630 to consider in curr_insn_transform.
35632 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35634 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
35635 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
35636 and emit_mode_inner.
35638 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35640 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
35641 argument to the overloaded builtin variants. Use the new flag to
35642 deprecate certain builtin variants.
35643 * config/s390/s390-builtin-types.def: Add new builtin types.
35644 * config/s390/s390-builtins.h: Support new flags field for
35645 overloaded builtins.
35646 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
35647 (s390_macro_to_expand): Enable vector float data type.
35648 (s390_cpu_cpp_builtins_internal): Indicate support of the new
35649 builtins by incrementing the __VEC__ version number.
35650 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
35652 (s390_resolve_overloaded_builtin): Emit error messages depending
35653 on the builtin flags.
35654 * config/s390/s390.c (s390_expand_builtin): Support additional
35655 flags argument. Change error message to match the messages
35656 emitted in s390-c.c.
35657 * config/s390/s390.md: New UNSPEC_* constants.
35658 (op_type): Add new instruction types.
35659 * config/s390/vecintrin.h: Add new builtins and test data class
35661 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
35662 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
35663 (VEC_INEXACT, VEC_NOINEXACT): New constants.
35664 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
35665 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
35666 ("vec_mergel<mode>"): V_HW -> VEC_HW.
35668 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
35669 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
35670 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
35671 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
35673 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
35674 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
35675 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
35676 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
35678 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
35679 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
35680 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
35681 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
35682 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
35683 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
35684 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
35686 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
35687 ("vec_scatter_element<V_HW_4:mode>_DI")
35688 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
35689 ("vec_fpint<mode>", "vflls")
35690 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
35691 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
35692 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
35693 ("*vec_cmphe<mode>_cc"): ... these.
35695 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
35696 mode constant instead of magic value.
35698 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35700 * config/s390/s390.c (s390_expand_vec_compare): Support other
35701 vector floating point modes than just V2DF.
35702 (s390_expand_vcond): Likewise.
35703 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
35704 (s390_cannot_change_mode_class): Prevent mode changes between TF
35705 and V1TF in vector registers.
35706 * config/s390/s390.md (DF, SF): New mode attributes.
35707 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
35708 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
35709 SFmode support for VRs.
35710 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
35712 (VFT, VF_HW): New mode iterators.
35713 (vw, sdx): New mode attributes.
35714 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
35715 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
35716 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
35717 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
35718 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
35719 also the new vector floating point modes. Renaming to ...
35721 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
35722 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
35723 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
35724 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
35725 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
35726 ("vec_unordered<mode>"): ... these.
35728 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
35729 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
35730 ("*vec_extendv2df"): New insn definitions.
35732 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35734 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
35735 ("mulditi3_2", "*muldi3_sign"): New patterns.
35736 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
35737 rename the pattern definition.
35739 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35741 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
35743 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
35745 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35747 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
35748 instruction if possible.
35749 * config/s390/vector.md (vec_halfnumelts): New mode
35751 ("*vec_vllezlf<mode>"): New pattern.
35753 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35755 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
35756 ("popcountv4si2", "popcountv2di2"): Rename to ...
35757 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
35758 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
35760 ("popcount<mode>2_vxe"): New pattern.
35762 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35764 * common/config/s390/s390-common.c (processor_flags_table): Add
35766 * config.gcc: Add arch12.
35767 * config/s390/driver-native.c (s390_host_detect_local_cpu):
35768 Default to arch12 for unknown CPU model numbers.
35769 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
35770 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
35771 PROCESSOR_max sanity check.
35772 * config/s390/s390-opts.h (enum processor_type): Add
35774 * config/s390/s390.c (processor_table): Add arch12.
35775 (s390_expand_builtin): Add check for B_VXE flag.
35776 (s390_issue_rate): Add PROCESSOR_ARCH12.
35777 (s390_get_sched_attrmask): Likewise.
35778 (s390_get_unit_mask): Likewise.
35779 (s390_sched_score): Enable z13 scheduling for arch12.
35780 (s390_sched_reorder): Likewise.
35781 (s390_sched_variable_issue): Likewise.
35782 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
35784 (s390_tune_attr): Use z13 scheduling also for arch12.
35785 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
35786 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
35787 (TARGET_VXE_P): New macros.
35788 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
35789 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
35790 * config/s390/s390.opt: Add arch12 as processor_type.
35792 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35794 * config/s390/s390.md
35795 ("fixuns_truncdddi2", "fixuns_trunctddi2")
35796 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
35797 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
35799 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
35800 Rename expanders to ...
35802 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
35803 ("fixuns_truncdddi2_emu"): ... these.
35805 ("fixuns_trunc<mode>si2_emu"): New expander.
35807 ("*fixuns_truncdfdi2_z13"): Rename to ...
35808 ("*fixuns_truncdfdi2_vx"): ... this.
35810 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35812 * config/s390/2964.md: Remove the single element vector compare
35813 instructions which are no longer used.
35814 * config/s390/s390.c (s390_select_ccmode): Remove handling of
35816 (s390_canonicalize_comparison): Remove handling of DFmode
35818 (s390_expand_vec_compare_scalar): Remove function.
35819 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
35820 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
35822 ("*cmp<mode>_ccs"): Add wfcdb instruction.
35824 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35826 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
35828 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
35829 will anyway by matched by mov<mode>_64dfp.
35831 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35833 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
35834 vlef/vstef. Add missing operand to vleif.
35836 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35838 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
35839 pair for all vector types with 64 bit elements.
35840 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
35841 * config/s390/vector.md (V_HW_64): ... here.
35842 (V_128_NOSINGLE): New mode iterator.
35843 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
35844 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
35845 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
35846 ("*vec_load_pairv2di"): Change to ...
35847 ("*vec_load_pair<mode>"): ... this one.
35849 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35851 * config/s390/constraints.md: Add comments.
35852 (jKK): Reject element sizes > 8 bytes.
35853 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
35855 * config/s390/s390.md: Add the s_operand checks formerly in
35856 s390_split_ok_p to various splitters where they are still
35858 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
35859 for 128 bit vectors. Plus two splitters.
35861 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35863 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
35866 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35869 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
35870 error if the boundary argument is not constant.
35872 2017-03-24 Jakub Jelinek <jakub@redhat.com>
35874 PR rtl-optimization/80112
35875 * loop-doloop.c (doloop_condition_get): Don't check condition
35876 if cmp isn't SET with IF_THEN_ELSE src.
35878 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35880 PR tree-optimization/80158
35881 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
35882 replacing a candidate statement, also replace it for the
35883 candidate's alternate interpretation.
35884 (replace_rhs_if_not_dup): Likewise.
35885 (replace_one_candidate): Likewise.
35887 2017-03-24 Richard Biener <rguenther@suse.de>
35889 PR tree-optimization/80167
35890 * graphite-isl-ast-to-gimple.c
35891 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
35893 (translate_isl_ast_to_gimple::get_rename): Likewise.
35895 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
35897 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
35898 handling of certain combinations of target options, including the
35899 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
35900 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
35902 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
35905 * config/arm/arm.md (*load_multiple): Add reload_completed to
35906 matching condition.
35908 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
35909 Richard Biener <rguenth@suse.de>
35911 PR tree-optimization/79908
35912 PR tree-optimization/80136
35913 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
35914 been cast away, gimplify_and_add suffices.
35916 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
35918 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
35920 2017-03-23 Richard Biener <rguenther@suse.de>
35922 PR tree-optimization/80032
35923 * gimplify.c (gimple_push_cleanup): Forced unconditional
35924 cleanups still have to go to the conditional_cleanups
35927 2017-03-22 Jakub Jelinek <jakub@redhat.com>
35929 PR tree-optimization/80072
35930 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
35932 (next_operand_entry_id): Change type to unsigned int.
35933 (sort_by_operand_rank): Make sure to return the right return value
35934 even if unsigned fields are bigger than INT_MAX.
35935 (struct oecount): Change cnt and id type to unsigned int.
35936 (oecount_hasher::equal): Formatting fix.
35937 (oecount_cmp): Make sure to return the right return value
35938 even if unsigned fields are bigger than INT_MAX.
35939 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
35942 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
35943 TREE_READONLY on result if writing it more than once.
35946 * doc/invoke.texi (-fsanitize=thread): Document that with
35947 -fnon-call-exceptions atomics are not able to throw
35951 * tsan.c: Include tree-eh.h.
35952 (instrument_builtin_call): Call maybe_clean_eh_stmt or
35953 maybe_clean_or_replace_eh_stmt where needed.
35954 (instrument_memory_accesses): Add cfg_changed argument.
35955 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
35956 if it returned true.
35957 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
35959 PR rtl-optimization/63191
35960 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
35961 wrapper function, moved the whole old content into ...
35962 (ix86_delegitimize_address_1): ... this. New inline function.
35963 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
35964 true as last argument instead of ix86_delegitimize_address.
35966 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35968 * config/aarch64/aarch64.c (generic_branch_cost): Copy
35969 cortexa57_branch_cost.
35971 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
35973 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
35975 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
35978 * doc/md.texi (Constraints): Document wA constraint.
35979 * config/rs6000/constraints.md (wA): New.
35980 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
35981 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
35982 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
35983 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
35985 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
35988 * gimplify.c (is_oacc_declared): New function.
35989 (oacc_default_clause): Use it to set default flags for acc declared
35990 variables inside parallel regions.
35991 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
35992 declared variables.
35993 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
35994 declare attribute to any decl as necessary.
35996 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
35999 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
36000 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
36001 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
36002 (arm_arch_lpae): This.
36003 * config/arm/arm.c (arm_arch7ve): Rename into ...
36004 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
36005 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
36008 2017-03-22 Martin Liska <mliska@suse.cz>
36011 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
36012 error message instead of an ICE.
36014 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36016 * doc/extend.texi (6.11 Additional Floating Types): Revise.
36018 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
36020 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
36022 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
36025 2017-03-21 Martin Sebor <msebor@redhat.com>
36027 * doc/extend.texi: Use "cannot" instead of "can't."
36028 * doc/hostconfig.texi: Same.
36029 * doc/install.texi: Same.
36030 * doc/invoke.texi: Same.
36031 * doc/loop.texi: Same.
36032 * doc/md.texi: Same.
36033 * doc/objc.texi: Same.
36034 * doc/rtl.texi: Same.
36035 * doc/tm.texi: Same.
36036 * doc/tm.texi.in: Same.
36037 * doc/trouble.texi: Same.
36039 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
36042 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
36043 (collect_checksum_attributes): Set it.
36044 (die_checksum_ordered): Use it.
36046 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36048 PR tree-optimization/79908
36049 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
36050 change: For a VA_ARG whose LHS has been cast away, use
36051 force_gimple_operand to construct the side effects.
36053 2017-03-21 David Malcolm <dmalcolm@redhat.com>
36055 PR translation/80001
36056 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
36057 more amenable to translation.
36058 (oacc_loop_auto_partitions): Likewise.
36060 2017-03-21 Marek Polacek <polacek@redhat.com>
36061 Martin Sebor <msebor@redhat.com>
36063 PR tree-optimization/80109
36064 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
36065 on INTEGRAL_TYPE_P.
36067 2017-03-21 Jakub Jelinek <jakub@redhat.com>
36068 Segher Boessenkool <segher@kernel.crashing.org>
36071 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
36072 check reg_used_between_p between insn and one of succ or succ2
36073 depending on if succ is artificial insn not inserted into insn
36076 2017-03-21 Martin Liska <mliska@suse.cz>
36078 PR gcov-profile/80081
36079 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
36080 * doc/gcc.texi: Include gcov-dump stuff.
36081 * doc/gcov-dump.texi: New file.
36083 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
36085 PR rtl-optimization/79150
36086 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
36087 conditional jump, if the jump is the last insn of the loop.
36089 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36090 Richard Biener <rguenth@suse.de>
36092 PR tree-optimization/79908
36093 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
36094 been cast away, use force_gimple_operand to construct the side
36097 2017-03-21 Martin Liska <mliska@suse.cz>
36099 PR libfortran/79956
36100 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
36103 2017-03-21 Brad Spengler <spender@grsecurity.net>
36106 * plugin.c (htab_hash_plugin): New function.
36107 (add_new_plugin): Use it and adjust.
36108 (parse_plugin_arg_opt): Adjust.
36109 (init_one_plugin): Likewise.
36111 2017-03-21 Richard Biener <rguenther@suse.de>
36113 PR tree-optimization/80032
36114 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
36115 if set force the cleanup to happen unconditionally.
36116 (gimplify_target_expr): Push inserted clobbers with force_uncond
36117 to avoid them being removed by control-dependent DCE.
36119 2017-03-21 Richard Biener <rguenther@suse.de>
36121 PR tree-optimization/80122
36122 * tree-inline.c (copy_bb): Do not expans va-arg packs or
36123 va_arg_pack_len when the inlined call stmt requires pack
36125 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
36127 2017-03-21 Jakub Jelinek <jakub@redhat.com>
36130 * tsan.c (instrument_builtin_call): If the memory model argument
36131 is not a constant, assume it is valid.
36134 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
36137 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36139 PR rtl-optimization/79910
36140 * combine.c (can_combine_p): Do not allow combining an I0 or I1
36141 if its dest is used by an insn before I2 (other than the combined
36142 insns themselves, which are properly handled already).
36144 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
36147 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36149 * combine.c (record_used_regs): New static function.
36150 (try_combine): Handle situations where there is an additional
36151 instruction between I2 and I3 which needs to have a LOG_LINK
36155 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36157 * combine.c (try_combine): Delete redundant i1 test. Call
36158 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36160 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
36163 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
36164 alternatives 13/14.
36166 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36168 PR tree-optimization/80054
36169 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
36170 the optimization if a PHI or any of its arguments is not dominated
36171 by the candidate's basis. Use gphi* rather than gimple* as
36173 (replace_profitable_candidates): Clean up a gimple* variable that
36174 should be a gphi* variable.
36176 2017-03-20 Martin Sebor <msebor@redhat.com>
36179 * doc/extend.texi (attribute constructor): Document present limitation.
36181 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
36184 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
36185 __POWER9_VECTOR__ #ifdef control, change template definition to
36186 use Power9-specific built-in function.
36187 (vec_any_eq): Likewise.
36188 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
36189 to control outcomes from this test.
36190 (vector_ae_<mode>p): For VEC_F modes, likewise.
36192 2017-03-20 Ian Lance Taylor <iant@google.com>
36194 * config/i386/i386.c (ix86_function_regparm): Save an extra
36195 register for -fsplit-stack with DECL_STATIC_CHAIN.
36197 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36200 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
36201 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
36203 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
36205 * config/riscv/riscv.c (riscv_print_operand): Use "fence
36207 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
36210 2017-03-20 Marek Polacek <polacek@redhat.com>
36213 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
36215 2017-03-20 Richard Biener <rguenther@suse.de>
36217 PR tree-optimization/80113
36218 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
36219 allocate extra SSA name for PHI def.
36220 (add_close_phis_to_outer_loops): Likewise.
36221 (add_close_phis_to_merge_points): Likewise.
36222 (copy_loop_close_phi_args): Likewise.
36223 (copy_cond_phi_nodes): Likewise.
36225 2017-03-20 Martin Liska <mliska@suse.cz>
36227 PR middle-end/79753
36228 * tree-chkp.c (chkp_build_returned_bound): Do not build
36229 returned bounds for a LHS that's not a BOUNDED_P type.
36231 2017-03-20 Martin Liska <mliska@suse.cz>
36235 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
36236 COMPLEX_CST and VECTOR_CST.
36238 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36241 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
36242 target operand. A new splitter adds the clobber statement in case
36243 the target operand is dead anyway.
36245 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
36247 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
36248 to age-old versions of binutils and glibc.
36250 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
36252 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
36254 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36256 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
36258 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
36260 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
36261 requirement for binutils 2.13.
36263 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
36265 * combine.c (try_combine): Delete redundant i1 test. Call
36266 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
36268 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
36270 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
36271 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
36273 <riscv64-*-elf>: Re-arrange section
36274 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
36275 <riscv32-*-linux>: Likewise.
36276 <riscv64-*-elf>: Likewise
36277 <riscv64-*-linux>: Likewise.
36279 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
36282 * aarch64.opt(verbose-cost-dump): Fix typo.
36284 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
36287 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
36288 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
36290 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
36292 * reload.c (find_reloads): When reloading a nonoffsettable address,
36293 use RELOAD_OTHER for it and its address reloads.
36295 PR rtl-optimization/79910
36296 * combine.c (record_used_regs): New static function.
36297 (try_combine): Handle situations where there is an additional
36298 instruction between I2 and I3 which needs to have a LOG_LINK
36301 2017-03-17 Jeff Law <law@redhat.com>
36303 PR tree-optimization/71437
36304 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
36305 conditional in the hash table first.
36306 (vrp_dom_walker::before_dom_children): Extract condition from
36307 ASSERT_EXPR. Record condition, its inverion and any implied
36308 conditions as well.
36310 2017-03-17 Marek Polacek <polacek@redhat.com>
36311 Markus Trippelsdorf <markus@trippelsdorf.de>
36313 PR tree-optimization/80079
36314 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
36317 2017-03-17 Richard Biener <rguenther@suse.de>
36319 PR middle-end/80075
36320 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
36321 Properly verify the LHS before the RHS possibly claims to be
36323 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
36326 2017-03-17 Martin Jambor <mjambor@suse.cz>
36328 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
36329 (List of -O2 options): Likewise.
36330 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
36333 2017-03-17 Tom de Vries <tom@codesourcery.com>
36335 * gcov-dump.c (print_usage): Print bug_report_url.
36337 2017-03-17 Richard Biener <rguenther@suse.de>
36339 PR middle-end/80050
36340 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
36341 (parser::peek): Likewise.
36343 2017-03-17 Richard Biener <rguenther@suse.de>
36345 PR tree-optimization/80048
36346 * sese.c (free_sese_info): Properly release rename_map and
36347 copied_bb_map elements.
36349 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
36351 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
36352 Add linked-list forward and backlinks. Insert on
36353 construction, remove on destruction.
36354 (class pass_store_merging): Add m_stores_head field.
36355 (pass_store_merging::terminate_and_process_all_chains):
36356 Iterate over m_stores_head list.
36357 (pass_store_merging::terminate_all_aliasing_chains):
36359 (pass_store_merging::execute): Check for debug stmts first.
36360 Push new chains onto the m_stores_head stack.
36362 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
36365 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
36366 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
36367 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
36369 2017-03-16 Jeff Law <law@redhat.com>
36371 PR tree-optimization/71437
36372 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
36373 member function. Implementation moved into after_dom_children
36374 member function and into the threader's thread_outgoing_edges
36376 (dom_opt_dom_walker::after_dom_children): Simplify by moving
36377 some code into new thread_outgoing_edges.
36378 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
36379 definition. Simplify marker handling (do it here). Assume we always
36380 have the available expression and the const/copies tables.
36381 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
36383 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
36384 * tree-vrp.c (equiv_stack): No longer file scoped.
36385 (vrp_dom_walker): New class.
36386 (vrp_dom_walker::before_dom_children): New member function.
36387 (vrp_dom_walker::after_dom_children): Likewise.
36388 (identify_jump_threads): Setup domwalker. Use it rather than
36389 walking edges in a random order by hand. Simplify setup/finalization.
36390 (finalize_jump_threads): Remove.
36391 (vrp_finalize): Do not call identify_jump_threads here.
36392 (execute_vrp): Do it here instead and call thread_through_all_blocks
36395 PR tree-optimization/71437
36396 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
36398 (simplify_stmt_for_jump_threading): Add basic_block argument. All
36400 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
36401 (dom_opt_dom_walker::thread_across_edge): Remove
36402 handle_dominating_asserts argument. All callers changed.
36403 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
36404 changes. Remove calls to lhs_of_dominating_assert. Other
36405 uses of handle_dominating_asserts turn into unconditional code
36406 (simplify_control_stmt_condition_1): Likewise.
36407 (simplify_control_stmt_condition): Likewise.
36408 (thread_through_normal_block, thread_across_edge): Likewise.
36409 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
36410 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
36411 object if it is not an SSA_NAME.
36412 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
36413 before calling into the VRP specific simplifiers.
36414 (identify_jump_threads): Remove handle_dominating_asserts
36417 2017-03-16 Jakub Jelinek <jakub@redhat.com>
36420 * tree-diagnostic.c (default_tree_printer): No longer static.
36421 * tree-diagnostic.h (default_tree_printer): New prototype.
36423 2017-03-16 Tamar Christina <tamar.christina@arm.com>
36425 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
36426 Change ins into fmov.
36428 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36430 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
36431 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
36432 Use h_con constraint for operand 1.
36433 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
36434 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
36436 2017-03-15 Jeff Law <law@redhat.com>
36438 PR tree-optimization/71437
36439 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
36440 (record_temporary_equivalences): Use it.
36442 PR tree-optimization/71437
36443 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
36444 tree-ssa-scopedtables.
36445 (lookup_avail_expr, build_and_record_new_cond): Likewise.
36446 (record_conditions, record_cond, vuse_eq): Likewise.
36447 (record_edge_info): Adjust to API tweak of record_conditions.
36448 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
36449 (record_temporary_equivalences, optimize_stmt): Likewise.
36450 (eliminate_redundant_computations): Likewise.
36451 (record_equivalences_from_stmt): Likewise.
36452 * tree-ssa-scopedtables.c: Include options.h and params.h.
36453 (vuse_eq): New function, moved from tree-ssa-dom.c
36454 (build_and_record_new_cond): Likewise.
36455 (record_conditions): Likewise. Accept vector of conditions rather
36456 than edge_equivalence structure for first argument.
36457 for the first argument.
36458 (avail_exprs_stack::lookup_avail_expr): New member function, moved
36459 from tree-ssa-dom.c.
36460 (avail_exprs_stack::record_cond): Likewise.
36461 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
36462 from tree-ssa-dom.c.
36463 (avail_exprs_stack): Add new member functions lookup_avail_expr
36465 (record_conditions): Declare.
36467 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
36470 * lra-constraints.c (process_alt_operands): Increase reject for
36471 reloading an input/output operand.
36473 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
36476 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
36477 insns to convert from signed/unsigned char/short to IEEE 128-bit
36479 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
36481 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
36484 * config/i386/i386.c (ix86_vector_duplicate_value): Create
36485 subreg of inner mode for values already in registers.
36487 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
36489 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
36490 iteration reg is used after the loop.
36492 2017-03-14 Martin Sebor <msebor@redhat.com>
36494 PR tree-optimization/79800
36495 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
36496 precision in negative-positive range.
36497 (format_floating): Call non-const overload with adjusted precision.
36499 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
36502 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
36505 2017-03-14 Martin Sebor <msebor@redhat.com>
36507 PR middle-end/80020
36508 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
36509 * builtins.def (aligned_alloc): Use it.
36512 * Makefile.in (GTFILES): Add calls.c.
36513 * calls.c: Include "gt-calls.h".
36515 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
36517 PR rtl-optimization/79728
36518 * regs.h (struct target_regs): New field
36519 x_contains_allocatable_regs_of_mode.
36520 (contains_allocatable_regs_of_mode): New macro.
36521 * reginfo.c (init_reg_sets_1): Initialize it, and change
36522 contains_reg_of_mode so it includes global regs as well.
36523 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
36524 rather than contains_regs_of_mode.
36526 2017-03-14 Martin Liska <mliska@suse.cz>
36528 * doc/invoke.texi: Document options that can't be combined with
36529 -fcheck-pointer-bounds.
36531 2017-03-14 Martin Liska <mliska@suse.cz>
36533 PR middle-end/79831
36534 * doc/invoke.texi (-Wchkp): Document the option.
36536 2017-03-14 Martin Liska <mliska@suse.cz>
36538 * Makefile.in: Install gcov-dump.
36540 2017-03-14 Martin Liska <mliska@suse.cz>
36542 * multiple_target.c (expand_target_clones): Bail out for
36543 an invalid attribute.
36545 2017-03-14 Richard Biener <rguenther@suse.de>
36547 * alias.c (struct alias_set_entry): Pack properly.
36548 * cfgloop.h (struct loop): Likewise.
36549 * cse.c (struct set): Likewise.
36550 * ipa-utils.c (struct searchc_env): Likewise.
36551 * loop-invariant.c (struct invariant): Likewise.
36552 * lra-remat.c (struct cand): Likewise.
36553 * recog.c (struct change_t): Likewise.
36554 * rtl.h (struct address_info): Likewise.
36555 * symbol-summary.h (function_summary): Likewise.
36556 * tree-loop-distribution.c (struct partition): Likewise.
36557 * tree-object-size.c (struct object_size_info): Likewise.
36558 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
36559 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
36560 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
36561 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
36562 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
36563 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
36564 (struct _stmt_vec_info): Likewise.
36566 2017-03-14 Martin Liska <mliska@suse.cz>
36569 * multiple_target.c (create_dispatcher_calls): Check that
36570 a target can create a function dispatcher.
36572 2017-03-14 Martin Liska <mliska@suse.cz>
36575 * multiple_target.c (expand_target_clones): Drop local.local
36576 flag for default implementation.
36578 2017-03-14 Richard Biener <rguenther@suse.de>
36580 PR tree-optimization/80030
36581 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
36583 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
36585 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
36586 gcc_fallthrough() instead of __attribute__((fallthrough));
36588 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36590 * doc/gcc.texi: Remove "up" link to (DIR).
36591 * doc/gccint.texi: Ditto.
36593 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
36595 * doc/install.texi (Specific) <avr>: Remove reference to
36598 2017-03-13 Jeff Law <law@redhat.com>
36600 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
36601 attribute rather than comments.
36603 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
36604 match_scratch operand is highest.
36606 2017-03-13 Martin Liska <mliska@suse.cz>
36608 PR middle-end/78339
36609 * ipa-pure-const.c (warn_function_noreturn): If the declarations
36610 is a CHKP clone, use original declaration.
36612 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36614 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
36615 (arc_conditional_register_usage): Use a different allocation order
36616 when optimizing for size.
36617 * common/config/arc/arc-common.c (arc_option_optimization_table):
36618 Section anchors default on when optimizing for size.
36620 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36622 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
36624 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36626 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
36627 * config/arc/arc.md (cpu_facility): Add cd variant.
36628 (*movqi_insn): Add code density variant.
36629 (*movhi_insn): Likewise.
36630 (*movqi_insn): Likewise.
36631 (*addsi3_mixed): Likewise.
36632 (subsi3_insn): Likewise.
36634 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36636 * config/arc/arc.md (movsi_cond_exec): Update constraint.
36638 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
36640 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
36641 expressions with MINUS and UNARY ops.
36643 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36646 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
36648 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
36649 between vec_select and vector argument.
36650 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
36651 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
36652 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
36653 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
36654 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
36655 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
36657 2017-03-13 Richard Biener <rguenther@suse.de>
36660 * params.def (vect-max-peeling-for-alignment): Fix typo.
36662 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36664 * doc/install.texi (Specific) <mips-*-*>: Remove description of
36665 issue that only occurred with binutils below 2.18.
36667 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36669 * doc/install.texi (Specific) <cris-axis-elf>: No longer
36670 refer to binutils 2.11/2.12 minimum.
36672 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
36674 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
36675 ftp.kernel.org and simplify binutils requirement.
36677 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
36679 * doc/invoke.texi (Warning Options): Fix spelling of link-time
36681 (Optimize Options): Ditto. Also remove redundancy.
36683 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36685 PR translation/79848
36686 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
36688 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
36689 to G_ to avoid double translation.
36691 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36693 PR translation/79923
36694 * auto-profile.c (get_combined_location): Convert leading
36695 character of diagnostics to lower case and remove trailing period.
36696 (read_profile): Likewise for various diagnostics.
36697 * config/arm/arm.c (arm_option_override): Remove trailing period
36698 from various diagnostics.
36699 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
36700 (msp430_expand_delay_cycles): Likewise.
36702 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36705 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
36706 full command-line argument, rather than just "str".
36707 (aarch64_validate_march): Likewise.
36708 (aarch64_validate_mtune): Likewise.
36710 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
36712 PR rtl-optimization/78911
36713 * lra-assigns.c (must_not_spill_p): New function.
36714 (spill_for): Use it.
36716 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36718 PR tree-optimization/79981
36719 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
36720 ATOMIC_COMPARE_EXCHANGE ifn result.
36721 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
36722 IFN_ATOMIC_COMPARE_EXCHANGE.
36724 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36727 * opts.c (parse_sanitizer_options): Add missing question mark to
36728 "did you mean" message.
36730 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36732 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
36734 (VMULEUH_UNS): Likewise.
36735 (VMULOUB_UNS): Likewise.
36736 (VMULOUH_UNS): Likewise.
36737 * config/rs6000/rs6000.c (builtin_function_type): Remove
36738 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
36740 2017-03-10 David Malcolm <dmalcolm@redhat.com>
36743 * read-rtl-function.c (function_reader::read_rtx_operand): Update
36744 x with result of extra_parsing_for_operand_code_0.
36745 (function_reader::extra_parsing_for_operand_code_0): Convert
36746 return type from void to rtx, returning x. When reading
36747 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
36748 larger size containing struct block_symbol.
36750 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
36752 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
36753 -mfloat128-hardware without -m64.
36755 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
36758 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
36759 entries to the case statement that marks unsigned arguments to
36760 overloaded functions.
36762 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
36764 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
36765 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
36767 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
36770 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
36771 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
36773 2017-03-10 Martin Liska <mliska@suse.cz>
36777 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
36778 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
36781 2017-03-10 Olivier Hainque <hainque@adacore.com>
36783 * tree-switch-conversion (array_value_type): Start by resetting
36784 candidate type to it's main variant.
36786 2017-03-10 Jakub Jelinek <jakub@redhat.com>
36788 PR rtl-optimization/79909
36789 * combine.c (try_combine): Use simplify_replace_rtx on individual
36790 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
36791 of the whole CALL_INSN_FUNCTION_USAGE.
36793 PR tree-optimization/79972
36794 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
36795 get_range_info on SSA_NAMEs. Formatting fixes.
36797 2017-03-10 Richard Biener <rguenther@suse.de>
36798 Jakub Jelinek <jakub@redhat.com>
36800 PR tree-optimization/77975
36801 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
36802 edge to be constant.
36803 (get_val_for): For constant x return it. Formatting fix.
36804 (loop_niter_by_eval): Avoid pointless looping if the next iteration
36805 would use the same bases as the current one.
36807 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
36809 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
36810 instead of vec_select for V1TImode.
36811 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
36813 (VSX_LE_128): Add V1TI to this mode iterator.
36814 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
36815 (*vsx_le_perm_store_<mode>): Likewise.
36816 (pre-reload splitter for VSX stores): Likewise.
36817 (post-reload splitter for VSX stores): Likewise.
36818 (*vsx_xxpermdi2_le_<mode>): Likewise.
36819 (*vsx_lxvd2x2_le_<mode>): Likewise.
36820 (*vsx_stxvd2x2_le_<mode>): Likewise.
36822 2017-03-09 Michael Eager <eager@eagercon.com>
36824 Correct failures with --enable-checking=yes,rtl.
36826 * config/microblaze/microblaze.c (microblaze_expand_shift):
36827 Replace GET_CODE test with CONST_INT_P and INTVAL test with
36828 test for const0_rtx.
36829 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
36830 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
36832 2017-03-09 Richard Biener <rguenther@suse.de>
36834 PR tree-optimization/79977
36835 * graphite-scop-detection.c (scop_detection::merge_sese):
36836 Handle the case of extra exits to blocks dominating the entry.
36838 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
36840 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
36843 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
36845 PR rtl-optimization/79949
36846 * lra-constraints.c (process_alt_operands): Check memory when
36847 trying to predict a cycle. Print about the overall increase.
36849 2017-03-09 Richard Biener <rguenther@suse.de>
36851 PR middle-end/79971
36852 * gimple-expr.c (useless_type_conversion_p): Preserve
36853 TYPE_SATURATING for fixed-point types.
36855 2017-03-09 Richard Biener <rguenther@suse.de>
36858 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
36859 alignment of BLKmode params.
36861 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
36864 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
36865 (VALL_NO_V2Q): Likewise.
36867 * config/aarch64/aarch64-simd.md
36868 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
36870 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
36871 VALL_NO_V2Q mode iterator.
36872 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
36874 2017-03-09 Martin Liska <mliska@suse.cz>
36876 PR tree-optimization/79631
36877 * tree-chkp-opt.c (chkp_is_constant_addr): Call
36878 tree_int_cst_sign_bit just for INTEGER constants.
36880 2017-03-09 Martin Liska <mliska@suse.cz>
36884 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
36887 2017-03-09 Marek Polacek <polacek@redhat.com>
36890 * tree.c (inchash::add_expr): Handle TREE_VEC.
36892 2017-03-09 Martin Liska <mliska@suse.cz>
36895 (chkp_narrow_size_and_offset): New function.
36896 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
36897 (void chkp_parse_bit_field_ref): New function.
36898 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
36899 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
36901 2017-03-09 Martin Liska <mliska@suse.cz>
36904 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
36905 (chkp_find_bounds_1): Remove gcc_unreachable.
36907 2017-03-09 Jakub Jelinek <jakub@redhat.com>
36910 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
36911 BUILT_IN_SYNC*, determine the access type from the size suffix and
36912 always build a MEM_REF with that type. Handle forgotten
36913 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
36916 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
36917 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
36918 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
36919 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
36920 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
36921 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
36922 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
36923 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
36924 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
36925 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
36926 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
36927 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
36928 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
36929 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
36930 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
36931 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
36932 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
36933 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
36934 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
36935 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
36936 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
36937 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
36938 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
36939 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
36940 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
36941 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
36942 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
36943 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
36944 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
36945 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
36946 definitions outside of __OPTIMIZE__ guarded section.
36949 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
36950 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
36951 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
36952 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
36955 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
36957 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
36958 ("vfenez<mode>"): Add missing constraints.
36960 2017-03-08 Martin Sebor <msebor@redhat.com>
36963 * config/nds32/nds32.c (nds32_option_override):
36964 Fix misspelled diagnostic.
36966 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36969 * gimplify.c (gimplify_omp_for): Replace index var in outer
36970 taskloop statement with an artificial variable and add
36971 OMP_CLAUSE_PRIVATE clause for it.
36973 2017-03-08 Richard Biener <rguenther@suse.de>
36975 PR tree-optimization/79955
36976 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
36977 for accesses that are completely outside of the variable.
36979 2017-03-08 Andrew Haley <aph@redhat.com>
36981 PR tree-optimization/79943
36982 * tree-ssa-loop-split.c (compute_new_first_bound): When
36983 calculating the new upper bound, (END-BEG) should be added, not
36986 2017-03-08 Jakub Jelinek <jakub@redhat.com>
36988 * config/avr/avr.md (setmemhi): Make sure match_dup
36989 operand number comes before match_scratch.
36991 2017-03-08 Richard Biener <rguenther@suse.de>
36993 PR tree-optimization/79920
36994 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
36995 with ncopies == 1 to ...
36996 (vect_transform_slp_perm_load): ... here. Properly compute
36997 all element loads by iterating VF times over the group. Do
36998 not handle ncopies (computed in a broken way) in
36999 vect_create_mask_and_perm.
37001 2017-03-08 Jakub Jelinek <jakub@redhat.com>
37004 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
37005 is a uniform vector, use uniform_vector_p return value instead of
37006 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
37008 2017-03-07 Marek Polacek <polacek@redhat.com>
37010 PR middle-end/79809
37011 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
37012 (alloca_call_type): Likewise.
37014 2017-03-07 Martin Liska <mliska@suse.cz>
37016 * gcov.c (process_args): Put comment to correct location.
37018 2017-03-07 Martin Liska <mliska@suse.cz>
37020 PR middle-end/68270
37021 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
37022 Use array_at_struct_end_p instead of DECL_CHAIN (field).
37023 (chkp_narrow_bounds_for_field): Likewise.
37024 (chkp_parse_array_and_component_ref): Pass one more argument to
37027 2017-03-07 Richard Biener <rguenther@suse.de>
37029 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
37032 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
37034 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
37035 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
37037 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37040 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
37041 to end of description.
37042 (PARAM_MAX_STORES_TO_MERGE): Likewise.
37044 2017-03-07 Jakub Jelinek <jakub@redhat.com>
37046 PR rtl-optimization/79901
37047 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
37049 (*avx512f_<code><mode>3<mask_name>): ... this.
37050 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
37051 iterator instead of VI8_AVX2_AVX512BW.
37053 PR rtl-optimization/79901
37054 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
37055 min/max expander, expand it using expand_vec_cond_expr.
37058 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
37061 2017-03-06 Jakub Jelinek <jakub@redhat.com>
37064 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
37065 to void * for PCH reasons.
37066 * dwarf2out.c (output_loc_operands, output_die): Cast
37067 v.val_vec.array to unsigned char *.
37069 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
37072 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
37075 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
37077 PR rtl-optimization/79571
37078 * lra-constraints.c (process_alt_operands): Calculate static
37079 reject and subtract it from overall when only addresses will be
37082 2017-03-06 Julia Koval <julia.koval@intel.com>
37085 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
37086 incoming stack boundary to 128 for 64-bit targets.
37088 2017-03-06 Richard Biener <rguenther@suse.de>
37090 PR tree-optimization/79894
37091 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
37092 to NULL after folding it.
37094 2017-03-06 Richard Biener <rguenther@suse.de>
37096 PR tree-optimization/79824
37097 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
37098 check disabling peeling for gaps.
37100 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
37102 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
37103 attributes): Document gettimeofday.
37105 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37107 * config/s390/s390.c (s390_option_override_internal): Set
37108 PARAM_MIN_VECT_LOOP_BOUND
37110 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
37112 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
37113 * config/s390/s390.md: Likewise.
37115 2017-03-06 Jakub Jelinek <jakub@redhat.com>
37118 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
37119 (<avx2_avx512>_perm<mode>): Rename to ...
37120 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
37122 (<avx512>_perm<mode>_mask): Rename to ...
37123 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
37125 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
37126 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
37127 instead of VI8F_256_512.
37128 (avx512f_perm<mode>): New define_expand.
37129 (avx512f_perm<mode>_mask): Likewise.
37130 (avx512f_perm<mode>_1<mask_name>): New define_insn.
37131 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
37133 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37135 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
37136 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
37138 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
37140 2017-03-06 Martin Liska <mliska@suse.cz>
37143 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
37144 when having a SSA NAME w/o VAR_DECL assigned to it.
37146 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37148 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
37149 msa_dpsub_<su>_d): Fix MODE for vec_select.
37151 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
37153 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
37155 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
37157 2017-03-06 Richard Biener <rguenther@suse.de>
37159 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
37160 * plugin.c (register_plugin_info): Likewise.
37161 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
37163 2017-03-05 Jakub Jelinek <jakub@redhat.com>
37165 * config/i386/sse.md (sse_storehps, sse_storelps,
37166 avx_<castmode><avxsizesuffix>_<castmode>,
37167 avx512f_<castmode><avxsizesuffix>_<castmode>,
37168 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
37169 in condition that at least one operand is not a MEM.
37171 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37173 PR middle-end/79805
37174 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
37175 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
37177 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
37178 gimple_call_nothrow_p flag based on whether original builtin can throw.
37179 If it can, emit following stmts on the fallthrough edge.
37180 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
37181 don't create new bb if inserting just debug stmts on the edge, try to
37182 insert them on the fallthru bb or just reset debug stmts.
37184 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
37187 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
37188 restore recog_data (including the operand rtxes inside it) around
37189 the call to get_insn_template.
37191 2017-03-03 Martin Sebor <msebor@redhat.com>
37193 PR tree-optimization/79699
37194 * context.c (context::~context): Free MPFR caches to avoid
37195 a memory leak on program exit.
37197 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
37199 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
37200 Use wide_int::ulow () instead of .elt (0).
37202 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37204 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
37205 (*pushxf): Limit oF constraint to 32bit targets and add oC
37206 constraint for 64bit targets.
37207 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
37208 (*pushdf): Change rmF constraint to rmC.
37210 2017-03-03 Martin Liska <mliska@suse.cz>
37212 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
37213 Remove unused variable.
37215 2017-03-03 Jakub Jelinek <jakub@redhat.com>
37218 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
37219 is a memory operand, increase num_memory.
37220 (ix86_expand_args_builtin): Likewise.
37222 2017-03-03 Jan Hubicka <jh@suse.cz>
37225 * ipa-devirt.c (maybe_record_node): Properly handle
37226 __cxa_pure_virtual visibility.
37228 2017-03-03 Martin Liska <mliska@suse.cz>
37230 PR tree-optimization/79803
37231 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
37233 (pass_loop_prefetch::execute): Disabled optimization if an
37234 assumption about L1 cache size is not met.
37236 2017-03-03 Martin Liska <mliska@suse.cz>
37238 PR rtl-optimization/79574
37239 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
37240 (hash_scan_set): Likewise.
37241 (dump_hash_table): Likewise.
37242 (hoist_code): Likewise.
37244 2017-03-03 Richard Biener <rguenther@suse.de>
37246 * fixed-value.c (fixed_from_string): Restore use of elt (1)
37247 in place of uhigh ().
37248 (fixed_convert_from_real): Likewise.
37250 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
37253 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
37255 2017-03-03 Richard Biener <rguenther@suse.de>
37257 PR middle-end/79818
37258 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
37259 TYPE_OVERFLOW_UNDEFINED check.
37261 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37263 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
37265 (vector_ae_<mode>_p): Likewise.
37266 (vector_nez_<mode>_p): Likewise.
37267 (vector_ne_v2di_p): Likewise.
37268 (vector_ae_v2di_p): Likewise.
37269 (vector_ne_<mode>_p): Likewise.
37270 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
37272 (vsx_tsqrt<mode>2_fe): Likewise.
37274 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
37277 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
37279 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37281 PR rtl-optimization/79780
37282 * cprop.c (one_cprop_pass): When second and further conditional trap
37283 in a single basic block is turned into an unconditional trap, turn it
37284 into a deleted note to avoid RTL verification failures.
37286 2017-03-02 Richard Biener <rguenther@suse.de>
37288 * fold-const.c (const_binop): Use ulow () instead of elt (0).
37290 2017-03-02 Richard Biener <rguenther@suse.de>
37292 PR tree-optimization/79345
37294 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
37295 param and abort the walk, returning -1 if it is hit.
37296 (walk_aliased_vdefs): Take a limit param and pass it on.
37297 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
37298 defaulting to 0 and return a signed int.
37299 * tree-ssa-uninit.c (struct check_defs_data): New struct.
37300 (check_defs): New helper.
37301 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
37302 about uninitialized memory.
37303 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
37304 bogus uninitialized warning.
37305 (fixed_convert_from_real): Likewise.
37307 2017-03-02 Bin Cheng <bin.cheng@arm.com>
37309 PR tree-optimization/66768
37310 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
37311 iv_use if base object can't be determined.
37313 2017-03-02 Jakub Jelinek <jakub@redhat.com>
37315 PR tree-optimization/79345
37316 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
37317 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
37318 (get_pattern_stats): Initialize it.
37319 * genemit.c (gen_expand): Verify match_scratch numbers come after
37320 match_operand/match_dup numbers.
37321 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
37322 match_scratch numbers.
37323 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
37325 * config/s390/s390.md (trunctdsd2): Likewise.
37327 2017-03-02 Richard Biener <rguenther@suse.de>
37329 * wide-int.h (wide_int_storage::operator=): Implement in terms
37332 2017-03-02 Richard Biener <rguenther@suse.de>
37334 PR tree-optimization/79777
37335 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
37336 the to insert expression to sth existing.
37338 2017-03-01 Martin Sebor <msebor@redhat.com>
37340 PR middle-end/79692
37341 * gimple-ssa-sprintf.c
37342 (directive::known_width_and_precision): New function.
37343 (format_integer): Use it.
37344 (get_mpfr_format_length): Consider the full range of precision
37345 when computing %g output with the # flag. Set the likely byte
37346 count to 3 rather than 1 when precision is indeterminate.
37347 (format_floating): Correct the lower bound of precision.
37349 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
37351 * doc/invoke.texi: Document default code model for 64-bit Linux.
37353 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
37356 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
37357 udiv rather than div since input pattern is unsigned.
37359 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
37361 * config/i386/i386.c (print_reg): Warn for values of
37362 unsupported size in integer register.
37364 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
37367 * config/rs6000/predicates.md (current_file_function_operand): Do
37368 not allow self calls to be local if the function is replaceable.
37370 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
37373 * config/rs6000/altivec.h (vec_ctz and others): Change the
37374 preprocessor macro that controls conditional compilation from
37375 _ARCH_PWR9 to __POWER9_VECTOR__.
37376 (vec_all_ne): Change parameterization of __altivec_scalar_pred
37377 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
37378 control (instead of _ARCH_PWR9 control) so that template
37379 definition uses power9-specific function.
37380 (vec_any_eq): Likewise.
37381 (vec_all_ne): Change macro definition to use a power9-specific
37382 expansion under #ifdef __POWER9_VECTOR__ control (instead of
37383 _ARCH_PWR9 control).
37384 (vec_any_eq) Likewise.
37385 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
37386 expansion for CMPNEF to remove support for xvcmpnesp instruction.
37387 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
37388 support for xvcmpnedp instruction.
37389 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
37390 macro expansion so that Power9 implementation of vec_all_ne does
37391 not use the AltiVec predicate framework.
37392 (VCMPNEH_P): Likewise.
37393 (VCMPNEW_P): Likewise.
37394 (VCMPNED_P): Likewise.
37395 (VCMPNEFP_P): Likewise.
37396 (VCMPNEDP_P): Likewise.
37397 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
37398 implementation of vec_any_eq to not use AltiVec predicate
37400 (VCMPAEH_P): Likewise.
37401 (VCMPAEW_P): Likewise.
37402 (VCMPAED_P): Likewise.
37403 (VCMPAEFP_P): Likewise.
37404 (VCMPAEDP_P): Likewise.
37405 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
37406 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
37407 not use the AltiVec predicate framework.
37408 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
37409 of vec_any_eq to not use AltiVec predicate framework.
37410 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
37411 support for predefined __POWER9_VECTOR__ macro to indicate that
37412 Power9 instruction selection is enabled.
37413 (altivec_overloaded_builtins): Remove extraneous
37414 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
37415 function argument types RS6000_BTI_bool_V16QI and
37416 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
37417 entry for overloaded function argument types RS6000_BTI_bool_V4SI
37418 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
37419 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
37420 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
37421 Power9 for implementations of vec_cmpne. Change the signature for
37422 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
37423 (representing vec_all_ne) to remove the previously described first
37424 argument of type RS6000_BTI_INTSI, as this was an artifact of
37425 reliance on the AltiVec predicate framework, which is no longer
37426 used in the implementation of these functions. Add
37427 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
37428 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
37429 since, unlike the AltiVec predicate framework implementation, we
37430 do not share function descriptors between vec_alle and vec_anyeq.
37431 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
37432 set of modes that receive special treatment even when
37433 TARGET_P9_VECTOR is true. The special treatment emits code that
37434 does not depend on Power9 instructions.
37435 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
37436 define_expand to not rely on AltiVec predicate framework.
37437 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37439 (vector_ne_v2di_p): Change this define_expand to not rely on
37440 AltiVec predicate framework.
37441 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
37443 (vector_ne_<mode>_p): Change this define_expand to not rely on
37444 AltiVec predicate framework.
37445 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
37447 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
37448 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
37449 define_insn pattern.
37450 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
37451 define_insn pattern because the xvcmpne<VSs>. instruction is not
37453 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
37454 instruction is not supported.
37456 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37458 * config/nvptx/nvptx.c: Include intl.h.
37460 2017-03-01 Martin Jambor <mjambor@suse.cz>
37463 * ipa-prop.h (ipa_bits): Removed field known.
37464 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
37465 to pointers. Adjusted their comments to warn about their sharing.
37466 (ipcp_transformation_summary): Change bits to a vector of pointers.
37467 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
37468 (ipa_get_ipa_bits_for_value): Declare.
37469 * tree-vrp.h (value_range): Mark as GTY((for_user)).
37470 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
37471 (ipa_bits_hash_table): Likewise.
37472 (ipa_vr_ggc_hash_traits): Likewise.
37473 (ipa_vr_hash_table): Likewise.
37474 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
37475 being pointers and vr_known being removed.
37476 (ipa_set_jf_unknown): Likewise.
37477 (ipa_get_ipa_bits_for_value): New function.
37478 (ipa_set_jfunc_bits): Likewise.
37479 (ipa_get_value_range): New overloaded functions.
37480 (ipa_set_jfunc_vr): Likewise.
37481 (ipa_compute_jump_functions_for_edge): Use the above functions to
37482 construct bits and vr parts of jump functions.
37483 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
37484 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37486 (ipcp_grow_transformations_if_necessary): Also allocate
37487 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
37489 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
37490 them. Fix too long lines.
37491 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
37492 vr_known being removed.
37493 (ipa_read_jump_function): Use new setter functions to construct bits
37494 and vr parts of jump functions or set them to NULL.
37495 (write_ipcp_transformation_info): Adjust for bits being pointers.
37496 (read_ipcp_transformation_info): Likewise.
37497 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
37499 Include gt-ipa-prop.h.
37500 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
37502 (ipcp_store_bits_results): Likewise.
37503 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
37504 Do not write to existing jump functions but use a temporary instead.
37506 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37509 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
37510 attempt to use its first operand as BIT_FIELD_REF base.
37512 2017-03-01 Richard Biener <rguenther@suse.de>
37514 PR middle-end/79721
37515 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
37516 interpolating formula in wrapping arithmetic.
37517 (chrec_apply): Convert chrec_evaluate return value to wanted type.
37519 2017-03-01 Jakub Jelinek <jakub@redhat.com>
37521 PR tree-optimization/79734
37522 * tree-vect-generic.c (expand_vector_condition): Optimize
37523 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
37524 Handle VEC_COND_EXPR where comparison has different inner width from
37525 type's inner width.
37527 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
37529 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
37530 markup, and similar issues. Remove @opindex entries for things
37531 that aren't options. Add missing -mmpy-option entries.
37533 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37535 PR tree-optimization/79737
37536 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
37537 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
37538 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
37539 instead of byte_size. Formatting fix.
37540 (shift_bytes_in_array_right): Formatting fix.
37542 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
37545 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
37546 condition on optimize for the leaf function test.
37548 2017-02-28 Martin Liska <mliska@suse.cz>
37551 * read-rtl-function.c (function_reader::handle_unknown_directive):
37552 Bail out when one uses -flto.
37554 2017-02-28 Martin Liska <mliska@suse.cz>
37556 * common.opt: Replace space with tabular for options of <number>
37558 * config/i386/i386.opt: Show <number> value for
37559 -mlarge-data-threshold.
37560 * opts.c (print_filtered_help): Do not display number in hexadecimal
37563 2017-02-28 Martin Liska <mliska@suse.cz>
37565 * common.opt: Fix --help=option -Q for options which are of
37568 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
37570 * config/i386/i386.c (print_reg): Error out for values
37571 of 8-bit size in invalid integer register.
37573 2017-02-28 Martin Sebor <msebor@redhat.com>
37575 PR tree-optimization/79691
37576 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
37578 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37581 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
37582 gcc_unreachable with output_operand_lossage.
37584 2017-02-28 Richard Biener <rguenther@suse.de>
37586 PR tree-optimization/79740
37587 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
37589 (visit_nary_op): Insert the nary into the hashtable if we
37590 pattern-matched sth.
37591 * tree-ssa-pre.c (eliminate_insert): Robustify.
37593 2017-02-28 Richard Biener <rguenther@suse.de>
37595 PR middle-end/79731
37596 * fold-const.c (decode_field_reference): Reject out-of-bound
37599 2017-02-28 Jakub Jelinek <jakub@redhat.com>
37601 * config/i386/i386.c: Include intl.h.
37602 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
37603 instead of just cond ? "..." : "...".
37604 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
37605 * coverage.c (read_counts_file): Likewise.
37606 * omp-offload.c: Include intl.h.
37607 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
37608 of just cond ? "..." : "...".
37609 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
37610 of just cond ? "..." : "...".
37612 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
37615 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
37617 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
37618 'tune for' CPU name.
37619 * config/arm/arm-cpu-data.h: Regenerated.
37621 2017-02-28 Richard Biener <rguenther@suse.de>
37623 PR tree-optimization/79732
37624 * tree-inline.c (expand_call_inline): Do not shadow var.
37626 2017-02-28 Richard Biener <rguenther@suse.de>
37628 PR tree-optimization/79723
37629 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
37630 address-space properly.
37632 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
37634 * doc/optinfo.texi (Optimization groups): Fix option used for
37636 * doc/invoke.texi (-fopt-info): Document "omp".
37637 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
37638 (OPTGROUP_ALL): Add OPTGROUP_OMP.
37639 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
37640 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
37641 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
37643 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
37645 * dumpfile.c (optgroup_options): Instead of "openmp", associate
37646 OPTGROUP_OMP with "omp".
37648 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
37651 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
37652 for arithmetic shift of unsigned V2DI.
37654 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
37656 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
37657 arc/linux.h headers.
37658 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
37659 (LINK_SPEC): Likewise.
37660 (ARC_TLS_EXTRA_START_SPEC): Likewise.
37661 (EXTRA_SPECS): Likewise.
37662 (STARTFILE_SPEC): Likewise.
37663 (ENDFILE_SPEC): Likewise.
37664 (LIB_SPEC): Likewise.
37665 (TARGET_SDATA_DEFAULT): Likewise.
37666 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
37667 (MULTILIB_DEFAULTS): Likewise.
37668 (DWARF2_UNWIND_INFO): Likewise.
37669 * config/arc/big.h: New file.
37670 * config/arc/elf.h: Likewise.
37671 * config/arc/linux.h: Likewise.
37672 * config/arc/t-uClibc: Remove.
37674 2017-02-27 Bin Cheng <bin.cheng@arm.com>
37676 PR tree-optimization/77536
37677 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
37678 (tree_transform_and_unroll_loop): Use above function to compute the
37679 estimated niter of unrolled loop and use it when scaling profile.
37680 Also use count info rather than frequency if it's non-zero.
37681 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
37682 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
37683 (vect_transform_loop): Call above function.
37685 2017-02-27 Richard Biener <rguenther@suse.de>
37687 PR tree-optimization/45397
37688 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
37689 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
37690 (visit_nary_op): Add pattern matching for CSEing sign-changed
37691 or truncated operations with wider ones.
37693 2017-02-27 Richard Biener <rguenther@suse.de>
37695 PR tree-optimization/79690
37696 * tree-vect-stmts.c (vectorizable_store): Use vector type
37697 built from the DR with address-space.
37699 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
37701 * doc/invoke.texi (Optimize Options): Refine the description
37702 of asan-use-after-return.
37704 2017-02-25 Alan Modra <amodra@gmail.com>
37706 PR rtl-optimization/79584
37707 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
37708 base, not ad->base_term, the reg within base. Remove assertion
37709 that ad->base == ad->base_term. Replace gen_int_mode using
37710 bogus mode with const0_rtx.
37712 2017-02-25 Jakub Jelinek <jakub@redhat.com>
37714 PR middle-end/79396
37715 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
37716 FMA_EXPR like tcc_binary or tcc_unary.
37718 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
37721 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
37723 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
37724 (output_loc_operands): Handle DW_OP_call_ref and
37725 DW_OP_GNU_variable_value.
37726 (struct variable_value_struct): New type.
37727 (struct variable_value_hasher): Likewise.
37728 (variable_value_hash): New variable.
37729 (string_types): Remove.
37730 (copy_loc_descr): New function.
37731 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
37732 (prepend_loc_descr_to_each): New function.
37733 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
37734 instead of add_loc_descr_to_each if the first argument is single
37735 location list and the second has multiple.
37736 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
37737 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
37738 when looking for variable value which doesn't have other location info.
37739 (loc_list_from_tree): Formatting fix.
37740 (gen_array_type_die): Simplify DW_AT_string_length handling.
37741 (adjust_string_types): Remove.
37742 (gen_subprogram_die): Don't call adjust_string_types nor test/set
37743 string_types. Call resolve_variable_values.
37744 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
37745 (resolve_addr_in_expr): Likewise. Add A argument.
37746 (copy_deref_exprloc): Remove deref argument. Adjust for the
37747 original expression being DW_OP_GNU_variable_value with optionally
37748 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
37749 optionally after it.
37750 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
37751 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
37752 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
37753 (variable_value_hasher::hash, variable_value_hasher::equal): New
37755 (resolve_variable_value_in_expr, resolve_variable_value,
37756 resolve_variable_values, note_variable_value_in_expr,
37757 note_variable_value): New functions.
37758 (dwarf2out_early_finish): Call note_variable_value on all toplevel
37761 2017-02-24 Jakub Jelinek <jakub@redhat.com>
37764 * opts.h (handle_generated_option): Add GENERATED_P argument.
37765 * opts-common.c (handle_option): Adjust function comment.
37766 (handle_generated_option): Add GENERATED_P argument, pass it to
37768 (control_warning_option): Pass false to handle_generated_option
37770 * opts.c (maybe_default_option): Pass true to handle_generated_option
37772 * optc-gen.awk: Likewise.
37774 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37776 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
37777 a REG, look at the REG it is a SUBREG of.
37778 (splitter for cmpeqsi_t): Ditto.
37780 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37782 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
37783 the special USEs with the pattern of the insn, not the insn itself.
37785 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
37788 * doc/invoke.texi: Document -mload-store-pairs.
37790 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37791 Sandra Loosemore <sandra@codesourcery.com>
37793 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
37794 argument isn't a CONST_INT.
37795 (nios2_alternate_compare_const): Assert op is a CONST_INT.
37796 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
37797 (nios2_validate_compare): Bypass alternate compare logic if *op2
37798 is not a CONST_INT.
37799 (ldstwm_operation_p): Return false if first_base is not a REG or
37800 if first_offset is not a CONST_INT.
37802 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37804 * config/cris/cris.md: Use correct operand in a define_peephole2.
37806 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37808 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
37810 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
37812 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
37813 this_insn if it is an INSN or JUMP_INSN.
37814 (force_offsettable): Look at base, not at addr.
37815 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
37816 on things that aren't necessarily CONST_INTs.
37818 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
37820 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
37821 -mfpmath=sse is the default also for x86-32 targets with SSE2
37822 instruction set when @option{-ffast-math} is enabled
37824 2017-02-24 Jeff Law <law@redhat.com>
37826 PR rtl-optimizatoin/79286
37827 * ira.c (update_equiv_regs): Drop may_trap_p exception to
37830 2017-02-24 Richard Biener <rguenther@suse.de>
37832 PR tree-optimization/79389
37833 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
37836 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
37838 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
37839 function comment to reflect reality.
37840 (loop_exits_before_overflow): Fix typo in function description.
37842 2017-02-24 Richard Biener <rguenther@suse.de>
37844 PR tree-optimization/79389
37845 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
37846 properly that a threading opportunity exists. Detect conditional
37847 copy/constant propagation opportunities.
37849 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
37851 * config/visium/visium.md (type): Add trap.
37852 (b): New mode attribute.
37853 (*btst): Rename into...
37854 (*btst<mode>): ...this and adjust.
37855 (*cbranchsi4_btst_insn): Rename into...
37856 (*cbranch<mode>4_btst_insn): ...this and adjust.
37857 (trap): New define_insn.
37859 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37861 PR tree-optimization/79389
37862 * ifcvt.c (struct noce_if_info): Add rev_cond field.
37863 (noce_reversed_cond_code): New function.
37864 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
37865 reversed_comparison_code. Formatting fix.
37866 (noce_try_store_flag): Test rev_cond != NULL in addition to
37867 reversed_comparison_code.
37868 (noce_try_store_flag_constants): Likewise.
37869 (noce_try_store_flag_mask): Likewise.
37870 (noce_try_addcc): Use rev_cond if non-NULL instead of
37871 reversed_comparison_code.
37872 (noce_try_cmove_arith): Likewise. Formatting fixes.
37873 (noce_try_minmax, noce_try_abs): Clear rev_cond.
37874 (noce_find_if_block): Initialize rev_cond.
37875 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
37876 instead of false as last argument never attempt to reverse it
37879 2017-02-23 Bin Cheng <bin.cheng@arm.com>
37881 PR tree-optimization/79663
37882 * tree-predcom.c (combine_chains): Process refs in reverse order
37883 only for ZERO length chains, and add explaining comment.
37885 2017-02-23 Jeff Law <law@redhat.com>
37887 PR tree-optimization/79578
37888 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
37889 in call to operand_equal_p.
37891 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
37894 * config/i386/cpuid.h: Fix another undefined behavior.
37896 2017-02-23 Richard Biener <rguenther@suse.de>
37898 PR tree-optimization/79683
37899 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
37900 vector types for data-refs.
37902 2017-02-23 Martin Liska <mliska@suse.cz>
37904 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
37906 2017-02-23 Jakub Jelinek <jakub@redhat.com>
37908 PR middle-end/79665
37909 * internal-fn.c (get_range_pos_neg): Moved to ...
37910 * tree.c (get_range_pos_neg): ... here. No longer static.
37911 * tree.h (get_range_pos_neg): New prototype.
37912 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
37913 are known to be in between 0 and signed maximum inclusive, try to
37914 expand both unsigned and signed divmod and use the cheaper one from
37917 2017-02-22 Jeff Law <law@redhat.com>
37919 PR tree-optimization/79578
37920 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
37921 to compare base operands.
37923 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
37926 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
37927 gpc_reg_operand instead of fpr_reg_operand.
37929 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
37931 * config/mips/mips.c (mips_return_in_memory): Force FP
37932 vector types to be returned in memory for o32 ABI.
37934 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37936 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
37937 instead of DW_TAG_member for static data member declarations and don't
37938 set no_linkage_name for static inline data members.
37939 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
37942 2017-02-22 Martin Liska <mliska@suse.cz>
37944 * doc/invoke.texi: Replace inequality signs with square brackets
37947 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
37950 * lra-constraints.c (simplify_operand_subreg): Handle
37951 WORD_REGISTER_OPERATIONS targets.
37953 2017-02-22 Jakub Jelinek <jakub@redhat.com>
37956 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
37957 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
37958 elimination by swapping fld*.
37960 2017-02-22 Richard Biener <rguenther@suse.de>
37962 PR tree-optimization/79673
37963 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
37964 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
37965 irrelevant address-space qualifiers and avoiding a
37966 ADDR_SPACE_CONVERT_EXPR from fold_convert.
37968 2017-02-22 Richard Biener <rguenther@suse.de>
37970 PR tree-optimization/79666
37971 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
37972 to not symbolically negate if that may introduce undefined
37975 2017-02-22 Martin Liska <mliska@suse.cz>
37978 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
37979 * data-streamer-out.c (streamer_write_gcov_count_stream):
37981 * value-prof.c (stream_out_histogram_value): Make assert more
37982 precise based on type of counter.
37984 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
37987 * config/i386/i386.md (standard_x87sse_constant_load splitter):
37988 Use nonimmediate_operand instead of memory_operand for operand 1.
37989 (float-extend standard_x87sse_constant_load splitter): Ditto.
37991 2017-02-21 Jeff Law <law@redhat.com>
37993 PR tree-optimization/79621
37994 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
37995 blocks with edges to themselves.
37997 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38000 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
38001 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
38002 Use gimple_call_builtin_p.
38005 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
38006 on temporarily removed DEBUG_INSNs.
38008 PR tree-optimization/79649
38009 * tree-loop-distribution.c (classify_partition): Give up on
38010 non-generic address space loads/stores.
38012 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
38014 * doc/loop.texi (Loop manipulation): Remove nonexistent
38015 tree_ssa_loop_version from the documentation.
38016 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
38018 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38021 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
38022 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
38023 * config/rs6000/rs6000.c: Include except.h.
38024 (rs6000_expand_split_stack_prologue): Call
38025 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
38027 2017-02-21 Martin Jambor <mjambor@suse.cz>
38030 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
38031 have been analyzed.
38033 2017-02-21 Martin Jambor <mjambor@suse.cz>
38035 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
38036 for backward compatibility only.
38037 * doc/invoke.texi (Option Summary): Remove all references to
38038 -fipa-cp-alignment.
38040 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
38044 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38046 * lra-constraints.c (curr_insn_transform): Handle
38047 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
38049 2017-02-21 Martin Liska <mliska@suse.cz>
38051 * config/i386/i386.opt: Replace -masm-dialect with -masm.
38053 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
38055 PR translation/79638
38056 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
38058 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
38061 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
38062 (arm_function_ok_for_sibcall): Return false for an indirect call by
38063 descriptor if all the argument registers are used.
38064 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
38065 alignment of the function.
38067 2017-02-21 Jakub Jelinek <jakub@redhat.com>
38069 PR tree-optimization/61441
38070 * simplify-rtx.c (simplify_const_unary_operation): For
38071 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
38072 the sNaN unmodified.
38074 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
38076 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
38077 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
38078 instead of SYSTEM_HEADER_DIR.
38080 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
38081 Martin Liška <mliska@suse.cz>
38083 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
38084 Fix typos and grammar, use active voice, and clarify.
38086 2017-02-20 Marek Polacek <polacek@redhat.com>
38088 PR middle-end/79537
38089 * gimplify.c (gimplify_expr): Handle unused *&&L;.
38092 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
38094 2017-02-20 Jakub Jelinek <jakub@redhat.com>
38097 * config/i386/i386.c (ix86_expand_builtin): Handle
38098 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
38099 ix86_builtins_isa[fcode].isa as a requirement of those
38100 flags and any other flag in the bitmask.
38101 (ix86_init_mmx_sse_builtins): Use 0 instead of
38102 ~OPTION_MASK_ISA_64BIT as mask.
38103 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
38104 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
38105 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
38106 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
38108 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38111 * lra-constraints.c (split_reg): Check requested split mode
38112 is supported by the register.
38114 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38116 * lra-constraints.c (simplify_operand_subreg): Remove early
38119 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38122 * lra-constraints.c (curr_insn_transform): Tighten condition
38123 for converting SUBREG reloads from OP_OUT to OP_INOUT.
38125 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
38128 * lra-constraints.c (curr_insn_transform): Handle
38129 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
38131 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
38134 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
38136 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
38138 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
38141 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
38144 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38146 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
38149 2017-02-18 Jakub Jelinek <jakub@redhat.com>
38151 * final.c (last_columnnum, override_columnnum): New variables.
38152 (final_start_function): Set last_columnnum, pass it to begin_prologue
38153 hook and pass 0 to dwarf2out_begin_prologue.
38154 (final_scan_insn): Update override_columnnum. Pass last_columnnum
38155 to source_line debug hook.
38156 (notice_source_line): Compute last_columnnum and for debug_column_info
38157 return true on column changes.
38158 * debug.h (struct gcc_debug_hooks): Add column argument to
38159 source_line and begin_prologue hooks.
38160 (debug_nothing_int_charstar_int_bool): Remove prototype.
38161 (debug_nothing_int_int_charstar,
38162 debug_nothing_int_int_charstar_int_bool): New prototypes.
38163 (dwarf2out_begin_prologue): Add column argument.
38164 * debug.c (do_nothing_debug_hooks): Adjust source_line and
38165 begin_prologue hooks.
38166 (debug_nothing_int_charstar_int_bool): Remove.
38167 (debug_nothing_int_int_charstar,
38168 debug_nothing_int_int_charstar_int_bool): New functions.
38169 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
38170 through to dwarf2out_source_line.
38171 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
38172 (dwarf2out_source_line): Add column argument, emit it if requested.
38173 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
38175 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38176 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
38177 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
38178 through to dwarf2out_begin_prologue.
38179 (vmsdbgout_source_line): Add column argument, pass it through to
38180 dwarf2out_source_line.
38181 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
38182 dbxout_source_line caller.
38183 (dbxout_source_line): Add column argument.
38185 * common.opt (gno-column-info, gcolumn-info): New options.
38186 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
38187 (check_die): Also test for multiple DW_AT_decl_column attributes.
38188 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
38189 DW_AT_decl_column if requested.
38190 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
38192 (gen_variable_die): Likewise.
38193 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
38194 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
38197 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
38198 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
38199 (ix86_handle_option): Handle OPT_m3dnowa.
38200 * doc/invoke.texi (-m3dnowa): Document.
38201 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
38202 -m3dnowa instead of -m3dnow -march=athlon.
38205 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
38206 instead of gcc_assert for K, r and R code checks. Formatting fixes.
38208 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
38211 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
38212 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
38213 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
38214 generator for vsx_xxpermdi_<mode>_be.
38215 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
38216 force big-endian semantics.
38217 (vsx_xxpermdi_<mode>_be): New define_expand with same
38218 implementation as previous version of vsx_xxpermdi_<mode>.
38220 2017-02-17 Jakub Jelinek <jakub@redhat.com>
38222 PR tree-optimization/79327
38223 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
38224 variable, its initialization and use.
38226 2017-02-17 Julia Koval <julia.koval@intel.com>
38228 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
38229 (OPTION_MASK_ISA_PKU_UNSET): New.
38230 (ix86_handle_option): Handle -mrdpid.
38231 * config/i386/cpuid.h (bit_RDPID): New.
38232 * config/i386/driver-i386.c (host_detect_local_cpu):
38233 Detect RDPID feature.
38234 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
38235 * config/i386/i386-c.c (ix86_target_macros_internal):
38237 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
38238 (ix86_valid_target_attribute_inner_p): Add "rdpid".
38239 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
38240 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
38241 * config/i386/i386.md (define_insn "rdpid"): New.
38242 * config/i386/i386.opt Add -mrdpid.
38243 * config/i386/immintrin.h (_rdpid_u32): New.
38245 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
38247 PR rtl-optimization/79541
38248 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
38249 instead of transforming it into USE.
38251 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
38253 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
38254 If HONOR_SNANS (SFmode) force the input to a register.
38255 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
38256 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
38257 an frsp or similar insn.
38259 2017-02-17 Martin Liska <mliska@suse.cz>
38261 PR rtl-optimization/79577
38262 * params.def (selsched-max-sched-times): Increase minimum to 1.
38264 2017-02-17 Martin Liska <mliska@suse.cz>
38266 PR rtl-optimization/79574
38267 * gcse.c (want_to_gcse_p): Prevent integer overflow.
38269 2017-02-17 Martin Liska <mliska@suse.cz>
38271 PR tree-optimization/79529
38272 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
38273 ssa_defined_default_def_p to handle cases which are implicitly
38275 * tree-ssa.c (ssa_defined_default_def_p): New function.
38276 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
38277 which are implicitly defined.
38278 * tree-ssa.h (ssa_defined_default_def_p): Declare.
38280 2017-02-17 Richard Biener <rguenther@suse.de>
38282 PR middle-end/79576
38283 * params.def (max-ssa-name-query-depth): Limit to 10.
38285 2017-02-17 Richard Biener <rguenther@suse.de>
38287 PR tree-optimization/79552
38288 * tree-ssa-structalias.c (visit_loadstore): Properly verify
38291 2017-02-17 Richard Biener <rguenther@suse.de>
38294 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
38296 2017-02-17 Marek Polacek <polacek@redhat.com>
38298 PR middle-end/79536
38299 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
38300 (fold_negate_expr): New wrapper.
38302 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
38304 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
38305 Correct terminology and de-emphasize pre-standard behavior.
38307 2017-02-16 Alan Modra <amodra@gmail.com>
38309 PR rtl-optimization/79286
38310 * ira.c (def_dominates_uses): New function.
38311 (update_equiv_regs): Don't create an equivalence for insns that
38312 may trap where the register def does not dominate the use.
38314 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
38316 PR rtl-optimization/78127
38317 * lra.c (lra): Call lra_eliminate before finish the loop after
38320 2017-02-16 Richard Biener <rguenther@suse.de>
38322 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
38324 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
38325 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
38326 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
38327 (isl_val_int_from_wi): New function.
38328 (extract_affine_gmp): Rename to ...
38329 (extract_affine_wi): ... this, take a widest_int.
38330 (extract_affine_int): Just wrap extract_affine_wi.
38331 (add_param_constraints): Use isl_val_int_from_wi.
38332 (add_loop_constraints): Likewise, and extract_affine_wi.
38334 2017-02-15 Jeff Law <law@redhat.com>
38336 PR middle-end/79521
38337 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
38338 ira_init_register_move_cost_if_necessary.
38340 2017-02-15 Martin Sebor <msebor@redhat.com>
38342 PR middle-end/32003
38343 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
38344 removed in a prior commit.
38346 2017-02-15 Bin Cheng <bin.cheng@arm.com>
38348 PR tree-optimization/79347
38349 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
38350 counters during peeling.
38352 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
38354 * Makefile.in (site.exp): Remove "set ISLVER".
38356 2017-02-15 Jakub Jelinek <jakub@redhat.com>
38359 * real.c (real_from_integer): Call real_convert even for decimal.
38361 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
38364 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
38366 2017-02-14 Andrew Pinski <apinski@cavium.com>
38368 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
38369 cores and change the partno/implementer to be correct.
38370 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
38371 the 'B" as the implementer.
38372 * config/aarch64/aarch64-tune.md: Regenerate.
38374 2017-02-14 Carl Love <cel@us.ibm.com>
38376 * config/rs6000/rs6000.c: Add case statement entry to make the
38377 xvcvuxdsp built-in argument unsigned.
38378 * config/rs6000/vsx.md: Fix the source and return operand types so they
38379 match the instruction definitions from the ISA document. Fix typo
38380 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
38383 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
38386 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
38387 member early_clobber_alts.
38388 * lra-lives.c (reg_early_clobber_p): New.
38389 (process_bb_lives): Use it.
38390 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
38391 (debug_operand_data): Initialize early_clobber_alts.
38392 (setup_operand_alternative): Set up early_clobber_alts.
38393 (collect_non_operand_hard_regs): Ditto. Pass early clobber
38394 alternatives to new_insn_reg.
38395 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
38397 (lra_update_insn_regno_info): Pass the new arg.
38399 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38401 PR middle-end/79505
38402 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
38403 (new_oacc_loop_raw): Don't clear already cleared fields.
38406 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
38407 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
38408 _mm512_prefetch_i64gather_ps): New inline functions and macros.
38410 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
38413 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
38415 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
38418 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
38419 the extra instruction to the right place to store 128-bit constant
38422 2017-02-14 Martin Sebor <msebor@redhat.com>
38424 PR middle-end/79448
38425 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
38426 warning for strings of unknown length.
38428 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
38430 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
38432 2017-02-14 Jeff Law <law@redhat.com>
38435 * ira-costs.c (scan_one_insn): Initialize register move costs
38436 for pseudos seen in USE/CLOBBER insns.
38438 PR tree-optimization/79095
38439 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
38440 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
38441 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
38442 if the operands are known to be not equal, then the resulting range
38444 (intersect_ranges): If the new range is ~[0,0] and the old range is
38445 wide, then prefer ~[0,0].
38446 * tree-vrp.c (overflow_comparison_p_1): New function.
38447 (overflow_comparison_p): New function.
38448 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
38449 if NAME is used in an overflow test.
38450 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
38451 overflow check that can be expressed as an equality test, then adjust
38452 ops to be that equality test.
38454 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
38456 * config/s390/s390-builtin-types.def: Remove flags argument.
38457 * config/s390/s390.c (s390_init_builtins): Likewise.
38459 2017-02-14 Martin Liska <mliska@suse.cz>
38461 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
38462 vector. Fix trailing white spaces.
38464 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
38466 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
38469 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38471 PR rtl-optimization/68664
38472 * config/arm/arm.c (arm_sched_can_speculate_insn):
38473 New function. Declare prototype.
38474 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38476 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38478 PR rtl-optimization/68664
38479 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
38481 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
38483 2017-02-14 Amit Pawar <amit.pawar@amd.com>
38485 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
38486 max skip bytes for function, loop and jump.
38488 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
38490 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
38491 ABS_EXPR for gimple dump.
38493 2017-02-14 Jakub Jelinek <jakub@redhat.com>
38496 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
38498 PR tree-optimization/79408
38499 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
38500 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
38501 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
38502 also if rhs1 is INTEGER_CST.
38504 2017-02-14 Richard Biener <rguenther@suse.de>
38506 PR middle-end/79432
38507 * tree-into-ssa.c (insert_phi_nodes): When the function can
38508 have abnormal edges rewrite SSA names with broken use-def
38509 dominance out of SSA and register them for PHI insertion.
38511 2017-02-13 Martin Sebor <msebor@redhat.com>
38513 PR middle-end/79496
38514 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
38515 clearing info.nowrite flag when snprintf size argument is a range.
38517 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38519 * cprop.c (cprop_jump): Add missing space in string literal.
38520 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
38521 (get_constraint_for_component_ref): Likewise.
38522 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
38523 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
38524 * lra-constraints.c (process_alt_operands): Likewise.
38525 * ipa-inline.c (inline_small_functions): Likewise.
38526 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
38527 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
38528 * trans-mem.c (diagnose_tm_1_op): Likewise.
38529 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
38530 (grid_parallel_clauses_gridifiable): Likewise.
38532 * config/nvptx/mkoffload.c (process): Add space in between
38535 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
38536 "MOD4_SSE_REGS" and "ALL_REGS".
38538 * spellcheck.c (test_data): Add , in between "foo" and "food".
38540 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38543 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
38544 boundary crossing check and subsequent code generation agree.
38546 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38548 * config/aarch64/aarch64.c (has_memory_op): Delete.
38549 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
38552 2017-02-13 Jakub Jelinek <jakub@redhat.com>
38554 PR rtl-optimization/79388
38555 PR rtl-optimization/79450
38556 * combine.c (distribute_notes): When removing TEM_INSN for which
38557 corresponding dest has last value recorded, invalidate that last
38560 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38562 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
38563 of explicit '@'. Add missing assembly comment marker on branch costs
38566 2017-02-13 Nathan Sidwell <nathan@acm.org>
38568 * gengtype-lex.l (<in_struct>): Add '/'.
38570 2017-02-13 Martin Liska <mliska@suse.cz>
38573 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
38575 2017-02-13 Richard Biener <rguenther@suse.de>
38577 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
38579 * configure: Re-generate.
38580 * config.in: Likewise.
38581 * graphite-dependences.c: Simplify as if
38582 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
38583 * graphite-isl-ast-to-gimple.c: Likewise.
38584 * graphite-optimize-isl.c: Likewise.
38585 * graphite-poly.c: Likewise.
38586 * graphite-sese-to-poly.c: Likewise.
38587 * graphite.h: Likewise.
38588 * toplev.c: Include isl/version.h and use isl_version () for
38589 printing the ISL version.
38590 * doc/install.texi: Update ISL requirement.
38592 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38594 * doc/standards.texi (Standards): Update reference to
38597 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
38599 * doc/extend.texi (Named Address Spaces): sourceware.org now
38601 * doc/install.texi (Binaries): Ditto.
38604 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38606 * doc/cpp.texi: Replace "stringify"/"stringification" with C
38607 standard terminology "stringize"/"stringizing" throughout.
38608 * doc/cppinternals.texi: Likewise.
38610 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
38612 * doc/extend.texi: Fix some spelling mistakes and typos.
38613 * doc/invoke.texi: Likewise.
38615 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38618 * params.def (inline-min-speedup) Change from 10 to 8.
38620 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38622 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
38625 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
38628 * ipa-inline-analysis.c (get_minimal_bb): New function.
38629 (record_modified): Use it.
38630 (remap_edge_change_prob): Handle also ancestor functions.
38632 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
38634 * doc/contrib.texi (Contributors): Remove broken link into
38635 the Mauve CVS repository.
38637 2017-02-11 Jakub Jelinek <jakub@redhat.com>
38639 PR middle-end/79454
38640 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
38641 result computation whenever lhs doesn't have vector mode, not
38642 just when it has BLKmode.
38644 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38646 * doc/makefile.texi (profiledbootstrap): Refer to the
38647 installation instructions only in textual form.
38649 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
38652 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
38654 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
38656 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
38657 (Specific): Update mingw-w64 reference.
38659 (Specific): Remove broken link to Renesas RX processor.
38661 2017-02-10 Richard Biener <rguenther@suse.de>
38663 * toplev.c (process_options): Do not mention obsolete graphite
38664 options when printing sorry message about missing graphite support.
38665 Mention -floop-nest-optimize.
38667 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
38669 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
38670 (vtst_p16): Likewise.
38671 (vtstq_p8): Likewise.
38672 (vtstq_p16): Likewise.
38674 (vtstq_p64): Likewise.
38675 * config/arm/arm_neon.h (vgetq_lane_p64): New.
38676 (vset_lane_p64): New.
38677 (vsetq_lane_p64): New.
38679 2017-02-10 Jakub Jelinek <jakub@redhat.com>
38681 PR tree-optimization/79411
38682 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
38683 stmt operands are SSA_NAMEs used in abnormal phis.
38684 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
38687 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38690 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
38693 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
38695 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
38697 2017-02-09 Jakub Jelinek <jakub@redhat.com>
38699 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
38703 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
38704 not arbitrary TREE_CONSTANT.
38707 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
38708 "omp declare target link" attribute unless is_global_var.
38709 * omp-offload.c (find_link_var_op): Likewise.
38711 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
38712 Chung-Lin Tang <cltang@codesourcery.com>
38714 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
38716 (gimplify_adjust_omp_clauses): Don't delete TILE.
38717 (gimplify_omp_for): Deal with TILE.
38718 * internal-fn.c (expand_GOACC_TILE): New function.
38719 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
38721 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
38722 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
38724 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
38725 avoid DIV for outermost collapse var.
38726 (expand_oacc_for): Insert tile element loop as needed. Adjust.
38727 Remove out of date comments, fix whitespace.
38728 * omp-general.c (omp_extract_for_data): Deal with tiling.
38729 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
38730 adjust OLF_DIM_BASE value.
38731 (struct omp_for_data): Add tiling field.
38732 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
38733 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
38734 for auto loops. Remove default auto determining, moved to
38735 oacc_loop_fixed_partitions.
38736 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
38737 stmts, add e_mask field.
38738 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
38739 (oacc_thread_numbers): Use oacc_dim_call.
38740 (oacc_xform_tile): New.
38741 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
38742 (finish_oacc_loop): Adjust for ifns vector.
38743 (oacc_loop_discover_walk): Append loop abstraction sites to list,
38744 add case for GOACC_TILE fns.
38745 (oacc_loop_xform_loop): Delete.
38746 (oacc_loop_process): Iterate over call list directly, and add
38747 handling for GOACC_TILE fns.
38748 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
38750 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
38751 vector partitioning to outer loops. Assign 2 partitions to loops
38752 when available. Add TILE handling.
38753 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
38754 (execite_oacc_device_lower): Process GOACC_TILE fns,
38755 ignore unknown specs.
38756 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
38757 * tree.c (omp_clause_num_ops): Adjust TILE ops.
38758 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
38760 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
38762 * configure.ac (ACX_BUGURL): Update.
38763 * configure: Regenerate.
38765 2017-02-09 Richard Biener <rguenther@suse.de>
38767 PR tree-optimization/69823
38768 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
38769 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
38771 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38773 * config/arc/arc-c.def: Add __NPS400__ definition.
38774 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
38775 (TARGET_NPS400): Define.
38777 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
38779 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
38781 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
38782 pointer, arch_info.
38783 (arc_cpu_types): Fill the arch_info field with a pointer into the
38784 arc_arch_types table.
38785 (arc_selected_cpu): Declare.
38786 * config/arc/arc.c (arc_selected_cpu): Make global.
38787 (arc_selected_arch): Delete.
38788 (arc_base_cpu): Delete.
38789 (arc_override_options): Remove references to deleted variables,
38790 update access to arch information.
38791 (ARC_OPT): Update access to arch information.
38792 (ARC_OPTX): Likewise.
38793 * config/arc/arc.h (arc_base_cpu): Remove declaration.
38794 (TARGET_ARC600): Update access to arch information.
38795 (TARGET_ARC601): Likewise.
38796 (TARGET_ARC700): Likewise.
38797 (TARGET_EM): Likewise.
38798 (TARGET_HS): Likewise.
38799 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
38802 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
38805 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
38806 condition/operands for integer GE/LE/GEU/LEU operations.
38808 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
38810 PR translation/79397
38811 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
38814 2017-02-08 Martin Jambor <mjambor@suse.cz>
38817 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
38818 whether allocation happened.
38819 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
38820 nothing was allocated.
38822 2017-02-08 Jakub Jelinek <jakub@redhat.com>
38824 PR tree-optimization/79408
38825 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
38826 constant, but SSA_NAME with a known integer range, use the minimum
38827 of that range instead of op1 to determine if modulo can be replaced
38828 with its first operand.
38830 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
38832 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
38834 2017-02-08 Richard Biener <rguenther@suse.de>
38836 PR tree-optimization/71824
38837 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
38838 Check all loops contained in the merged region.
38840 2017-02-07 Andrew Pinski <apinski@cavium.com>
38842 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
38844 2017-02-07 Andrew Pinski <apinski@cavium.com>
38846 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
38847 (thunderxt88): Likewise.
38848 (thunderxt81): Disable LSE and change v8.1 to v8.
38849 (thunderxt83): Likewise.
38851 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38852 Richard Biener <rguenther@suse.de>
38854 PR middle-end/79399
38855 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
38856 type from int to size_t.
38857 * ira-costs.c (struct_costs_size): Change type from int to size_t.
38859 2017-02-07 Jakub Jelinek <jakub@redhat.com>
38861 PR rtl-optimization/79386
38862 * cprop.c (bypass_conditional_jumps): Initialize
38863 bypass_last_basic_block already before splitting bbs after
38864 unconditional traps...
38865 (bypass_conditional_jumps): ... rather than here.
38868 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
38869 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
38870 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
38871 fix -masm=intel patterns.
38873 2017-02-07 Richard Biener <rguenther@suse.de>
38875 PR tree-optimization/79256
38876 PR middle-end/79278
38877 * builtins.c (get_object_alignment_2): Use min_align_of_type
38878 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
38879 and ADJUST_FIELD_ALIGN.
38881 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
38883 * doc/tm.texi: Regenerate.
38884 * stor-layout.c (layout_decl): Adjust.
38885 (update_alignment_for_field): Likewise.
38886 (place_field): Likewise.
38887 (min_align_of_type): Likewise.
38888 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
38889 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
38890 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
38891 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
38892 * config/frv/frv.c (frv_adjust_field_align): Likewise.
38893 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
38894 * config/i386/i386.c (x86_field_alignment): Likewise.
38895 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
38896 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
38897 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
38898 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
38899 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
38900 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
38904 2017-01-30 Richard Biener <rguenther@suse.de>
38906 PR tree-optimization/79256
38907 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
38908 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
38911 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
38913 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
38914 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
38915 builtins to SImode and emit a zero-extend, if necessary.
38917 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38919 * docs/invoke.texi (RISC-V Options): Alphabetize.
38921 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38923 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
38926 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
38928 * config/riscv/riscv.c: New file.
38929 * common/config/riscv/riscv-common.c: Likewise.
38930 * config.gcc: Likewise.
38931 * config/riscv/constraints.md: Likewise.
38932 * config/riscv/elf.h: Likewise.
38933 * config/riscv/generic.md: Likewise.
38934 * config/riscv/linux.h: Likewise.
38935 * config/riscv/multilib-generator: Likewise.
38936 * config/riscv/peephole.md: Likewise.
38937 * config/riscv/pic.md: Likewise.
38938 * config/riscv/predicates.md: Likewise.
38939 * config/riscv/riscv-builtins.c: Likewise.
38940 * config/riscv/riscv-c.c: Likewise.
38941 * config/riscv/riscv-ftypes.def: Likewise.
38942 * config/riscv/riscv-modes.def: Likewise.
38943 * config/riscv/riscv-opts.h: Likewise.
38944 * config/riscv/riscv-protos.h: Likewise.
38945 * config/riscv/riscv.h: Likewise.
38946 * config/riscv/riscv.md: Likewise.
38947 * config/riscv/riscv.opt: Likewise.
38948 * config/riscv/sync.md: Likewise.
38949 * config/riscv/t-elf-multilib: Likewise.
38950 * config/riscv/t-linux: Likewise.
38951 * config/riscv/t-linux-multilib: Likewise.
38952 * config/riscv/t-riscv: Likewise.
38953 * configure.ac: Likewise.
38954 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
38955 Waterman as RISC-V maintainers.
38956 * doc/install.texi: Add RISC-V entries.
38957 * doc/invoke.texi: Add RISC-V options section.
38958 * doc/md.texi: Add RISC-V constraints section.
38959 * configure: Regenerated.
38961 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
38964 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
38965 false values to be constant vectors with all 0 or all 1 bits set.
38966 (vcondu<mode><mode>): Likewise.
38967 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
38969 (fpmask_comparison_operator): Update comment.
38970 (vecint_comparison_operator): New predicate.
38971 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
38972 vector conditionals when the true and false values are constant
38973 vectors with all 0 bits or all 1 bits set.
38975 2017-02-06 Martin Sebor <msebor@redhat.com>
38977 PR tree-optimization/79376
38978 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
38980 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
38982 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
38983 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
38984 to simplify split condition.
38986 2017-02-06 Jakub Jelinek <jakub@redhat.com>
38988 * omp-expand.c (oxpand_omp_atomic_fetch_op,
38989 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
38992 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
38994 PR rtl-optimization/68664
38995 * target.def (can_speculate_insn): New hook.
38996 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
38997 * doc/tm.texi: Regenerate.
38998 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
38999 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
39000 (rs6000_sched_can_speculate_insn): New function.
39002 2017-02-06 Jakub Jelinek <jakub@redhat.com>
39004 PR tree-optimization/79284
39005 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
39006 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
39007 vectorizable_mask_load_store, vectorizable_operation,
39008 vect_is_simple_cond, get_same_sized_vectype): Use it instead
39009 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
39010 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
39011 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
39012 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
39013 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
39014 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
39015 is_gimple_assign (stmt). Replace another such test with
39016 is_gimple_assign (stmt).
39018 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
39021 * config/avr/avr.c (rtl-iter.h): Include it.
39022 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
39023 (avr_legitimate_combined_insn): ...and implementation.
39025 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
39027 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
39028 * config/s390/s390.c (s390_const_operand_ok)
39029 (s390_canonicalize_comparison, s390_extract_part)
39030 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
39031 (s390_contiguous_bitmask_p, s390_rtx_costs)
39032 (legitimize_pic_address): Likewise.
39033 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
39034 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
39035 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
39036 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
39037 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
39039 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
39041 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
39042 REGNO($0) == REGNO($1).
39044 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39046 * config/s390/linux.h(SIZE_TYPE): Add comment.
39048 2017-02-06 Julian Brown <julian@codesourcery.com>
39049 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
39050 Virendra Pathak <virendra.pathak@broadcom.com>
39052 * config/aarch64/aarch64-cores.def: Change the scheduler
39054 * config/aarch64/aarch64.md: Include thunderx2t99.md.
39055 * config/aarch64/thunderx2t99.md: New file.
39057 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39059 * doc/standards.texi (Go Language): Update link to language
39062 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
39064 * tree-eh.c (lower_resx): Sanitize profile.
39065 (cleanup_empty_eh_move_lp): Likewise.
39067 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
39070 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
39072 * cfgloopmanip.h (loop_version): Update prototype.
39073 * modulo-sched.c (sms_schedule): Update call of loop_version.
39074 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
39075 * tree-parloops.c (gen_parallel_loop): Likewise.
39076 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
39077 * tree-ssa-loop-split.c (split_loop): Likewise.
39078 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
39079 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
39081 2017-02-05 Martin Liska <mliska@suse.cz>
39084 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
39086 (print_operand_address): Initialize a struct to zero.
39088 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39090 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
39091 garbage collector only in textual form.
39093 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
39095 * doc/extend.texi (x86 specific memory model extensions for
39096 transactional memory): Simplify a phrase.
39098 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
39101 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
39102 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
39103 (atomic_storedi_1): Likewise.
39105 2017-02-04 Jakub Jelinek <jakub@redhat.com>
39107 PR tree-optimization/79338
39108 * tree-parloops.c (gather_scalar_reductions): Don't call
39109 vect_analyze_loop_form for loop->inner before destroying loop's
39112 2017-02-03 Martin Sebor <msebor@redhat.com>
39114 PR tree-optimization/79327
39115 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
39116 when precision has resulted in leading zeros.
39117 (format_integer): Adjust the likely counter to assume an unknown
39118 argument that may be zero is non-zero.
39120 2017-02-03 Jason Merrill <jason@redhat.com>
39123 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
39124 avoid copying non-taken branch.
39126 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39128 PR tree-optimization/79340
39129 * tree-vect-loop.c (vectorizable_reduction): Release
39130 vec_defs elements after safe_splicing them into other vectors.
39133 PR tree-optimization/79327
39134 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
39135 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
39137 (format_integer): Use wide_int_to_tree instead of build_int_cst
39138 + to_?hwi. If argmin is NULL, just set argmin and argmax to
39139 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
39140 of shortest and longest sequence.
39142 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
39144 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
39145 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
39147 2017-02-03 Walter Lee <walt@tilera.com>
39150 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
39151 after initial stackframe link reg save.
39152 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
39154 2017-02-03 Jakub Jelinek <jakub@redhat.com>
39157 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
39158 wu for stxssp alternative.
39160 2017-02-03 Martin Sebor <msebor@redhat.com>
39162 PR tree-optimization/79352
39163 * gimple-fold.c (get_range_strlen): Add argument.
39164 (get_range_strlen): Change return type to bool.
39165 (get_maxval_strlen): Pass in a dummy argument.
39166 * gimple-fold.h (get_range_strlen): Change return type to bool.
39167 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
39168 * tree.h (array_at_struct_end_p): Add argument.
39169 * tree.c (array_at_struct_end_p): Handle it.
39171 2017-02-03 Martin Liska <mliska@suse.cz>
39174 * multiple_target.c (create_dispatcher_calls): Redirect edge
39175 from a caller of a dispatcher.
39176 (expand_target_clones): Make the clones local.
39177 (ipa_target_clone): Do both target clones and resolvers.
39178 (ipa_dispatcher_calls): Remove the pass.
39179 (pass_dispatcher_calls::gate): Likewise.
39180 (make_pass_dispatcher_calls): Likewise.
39181 * passes.def (pass_target_clone): Put as very first IPA early
39184 2017-02-03 Martin Liska <mliska@suse.cz>
39186 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
39187 in case of a function with ifunc attribute.
39189 2017-02-03 Martin Liska <mliska@suse.cz>
39191 * cgraph.c (cgraph_node::dump): Dump function version info.
39192 * symtab.c (symtab_node::dump_base): Add missing new line.
39194 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39196 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
39197 (ifcombine_ifandif): Use it.
39199 2017-02-03 Martin Liska <mliska@suse.cz>
39201 * doc/invoke.texi: Document default value for
39202 use-after-scope-direct-emission-threshold.
39204 2017-02-03 Martin Liska <mliska@suse.cz>
39206 PR tree-optimization/79339
39207 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
39208 (format_floating): Likewise.
39210 2017-02-03 Martin Liska <mliska@suse.cz>
39213 * ipa-prop.c (ipa_node_params_t::insert): Remove current
39215 (ipa_node_params_t::remove): Likewise.
39216 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
39217 initialization from removed ipa_node_params_t::insert.
39218 (ipa_node_params::~ipa_node_params): Move from removed
39219 ipa_node_params_t::release.
39220 * symbol-summary.h (symbol_summary::m_released): New member.
39221 Do not release a summary twice. Do not allow to call finalizer
39222 for types of a summary that live in GGC memory.
39224 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
39226 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
39229 2017-02-02 Martin Sebor <msebor@redhat.com>
39231 PR middle-end/79275
39232 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
39233 (format_string): Tighten up the range of output for non-constant
39234 strings and correct the expected range for wide non-constant strings.
39236 2017-02-02 Martin Sebor <msebor@redhat.com>
39238 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
39240 PR middle-end/32003
39241 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
39243 (-fdump-tree-@var): Add to index and document how to come up
39244 with pass-specific option and dump file names.
39245 (-fdump-passes): Clarify where to look for output.
39247 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
39249 PR middle-end/77445
39250 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
39251 statistics of the analyzed path; allow threading for speed when
39252 any of BBs along the path are optimized for speed.
39254 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
39256 PR middle-end/78468
39257 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
39258 settings of the virtual registers.
39261 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
39263 * explow.c (get_dynamic_stack_size): Take known alignment of stack
39264 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
39267 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39269 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
39270 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
39272 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39274 * config/s390/s390.md: Add missing comments with the expanded
39276 * config/s390/vector.md: Likewise.
39277 * config/s390/vx-builtins.md: Likewise.
39279 2017-02-02 Jakub Jelinek <jakub@redhat.com>
39282 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
39283 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
39284 conditions on a single line.
39286 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39288 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39289 __S390_VX__ to __VX__.
39291 2017-02-01 Andrew Pinski <apinski@cavium.com>
39293 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
39294 stmt_info to record_stmt_cost.
39295 (vect_get_known_peeling_cost): Pass stmt_info if known to
39297 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
39298 cpu_vector_cost field into
39299 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
39300 field into vec_int_stmt_cost and vec_fp_stmt_cost.
39301 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
39302 splitting of scalar_stmt_cost and vec_stmt_cost.
39303 (thunderx_vector_cost): Likewise.
39304 (cortexa57_vector_cost): LIkewise.
39305 (exynosm1_vector_cost): Likewise.
39306 (xgene1_vector_cost): Likewise.
39307 (thunderx2t99_vector_cost): Improve after the splitting of the two
39309 (aarch64_builtin_vectorization_cost): Update for the splitting of
39310 scalar_stmt_cost and vec_stmt_cost.
39312 2017-02-01 Torvald Riegel <triegel@redhat.com>
39313 Richard Henderson <rth@redhat.com>
39315 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
39316 conditional on existance of a fast atomic load.
39317 * optabs-query.c (can_atomic_load_p): New function.
39318 * optabs-query.h (can_atomic_load_p): Declare it.
39319 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
39320 no fast atomic load is available for the particular size of access.
39321 (expand_atomic_compare_and_swap): Likewise.
39322 (expand_atomic_load): Likewise.
39323 (expand_atomic_store): Likewise.
39324 (expand_atomic_fetch_op): Likewise.
39325 * testsuite/lib/target-supports.exp
39326 (check_effective_target_sync_int_128): Remove x86 because it provides
39327 no fast atomic load.
39328 (check_effective_target_sync_int_128_runtime): Likewise.
39330 2017-02-01 Richard Biener <rguenther@suse.de>
39332 * graphite.c: Include tree-vectorizer.h for find_loop_location.
39333 (graphite_transform_loops): Provide opt-info for optimized nests.
39334 * tree-parloop.c (parallelize_loops): Provide opt-info for
39335 parallelized loops.
39337 2017-02-01 Richard Biener <rguenther@suse.de>
39339 PR middle-end/79315
39340 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
39341 was not set before.
39343 2017-02-01 Richard Biener <rguenther@suse.de>
39345 PR tree-optimization/71824
39346 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
39347 Verify the loops are valid in the merged SESE region.
39348 (scop_detection::can_represent_loop_1): Check analyzing the
39349 evolution of the number of iterations in the region succeeds.
39351 2017-01-31 Ian Lance Taylor <iant@golang.org>
39353 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
39354 REG_ARGS_SIZE note to 32-bit push insns and call insn.
39356 2017-01-31 David Malcolm <dmalcolm@redhat.com>
39358 PR preprocessor/79210
39359 * input.c (get_substring_ranges_for_loc): Replace line_width
39360 assertion with error-handling.
39362 2017-01-31 Richard Biener <rguenther@suse.de>
39364 PR tree-optimization/77318
39365 * graphite-sese-to-poly.c (extract_affine): Fix assert.
39366 (create_pw_aff_from_tree): Take loop parameter.
39367 (add_condition_to_pbb): Pass loop of the condition to
39368 create_pw_aff_from_tree.
39370 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39372 * config/s390/s390.c (s390_asan_shadow_offset): New function.
39373 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
39375 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
39379 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
39381 (convert_int_to_float128): Likewise.
39382 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
39383 (convert_int_to_float128): Likewise.
39384 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
39385 (UNSPEC_IEEE128_CONVERT): Likewise.
39386 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
39387 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
39388 Use local variables for IBM extended format.
39389 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
39390 (fix_trunc<mode>si2_fprs): Likewise.
39391 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
39392 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
39393 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
39394 to know that we can now have integers of all sizes in vector
39396 (fix<uns>_<mode>di2_hw): Likewise.
39397 (float<uns>_<mode>si2_hw): Likewise.
39398 (fix_<mode>si2_hw): Likewise.
39399 (fixuns_<mode>si2_hw): Likewise.
39400 (float<uns>_<mode>di2_hw): Likewise.
39401 (float_<mode>di2_hw): Likewise.
39402 (float_<mode>si2_hw): Likewise.
39403 (floatuns_<mode>di2_hw): Likewise.
39404 (floatuns_<mode>si2_hw): Likewise.
39405 (xscvqp<su>wz_<mode>): Delete, no longer used.
39406 (xscvqp<su>dz_<mode>): Likewise.
39407 (xscv<su>dqp_<mode>): Likewise.
39408 (ieee128_mfvsrd_64bit): Likewise.
39409 (ieee128_mfvsrd_32bit): Likewise.
39410 (ieee128_mfvsrwz): Likewise.
39411 (ieee128_mtvsrw): Likewise.
39412 (ieee128_mtvsrd_64bit): Likewise.
39413 (ieee128_mtvsrd_32bit): Likewise.
39415 2017-01-31 Martin Liska <mliska@suse.cz>
39418 * ipa-prop.c (ipa_free_all_node_params): Call release method
39419 instead of ~sumbol_summary to not to trigger double times
39422 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
39424 PR tree-optimization/71691
39425 * bitmap.h (class auto_bitmap): New.
39426 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
39427 is_maybe_undefined instead of ssa_undefined_value_p.
39429 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
39431 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
39432 __S390_ARCH_LEVEL__ to __ARCH__.
39434 2017-01-31 Jakub Jelinek <jakub@redhat.com>
39436 PR tree-optimization/79267
39437 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
39438 if should_remove_lhs_p is true.
39440 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
39443 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
39444 (add_alignment_attribute): New.
39445 (base_type_die): Add alignment attribute.
39446 (subrange_type_die): Likewise.
39447 (modified_type_die): Likewise.
39448 (gen_array_type_die): Likewise.
39449 (gen_descr_array_type_die: Likewise.
39450 (gen_enumeration_type_die): Likewise.
39451 (gen_subprogram_die): Likewise.
39452 (gen_variable_die): Likewise.
39453 (gen_field_die): Likewise.
39454 (gen_ptr_to_mbr_type_die): Likewise.
39455 (gen_struct_or_union_type_die): Likewise.
39456 (gen_subroutine_type_die): Likewise.
39457 (gen_typedef_die): Likewise.
39458 (base_type_cmp): Compare alignment attribute.
39460 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
39463 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
39464 (setb_unsigned) New pattern for setb with CCUNS.
39465 * config/rs6000/rs6000.c (expand_block_compare): Use a different
39466 subfc./subfe sequence to avoid overflow problems. Generate a
39467 shorter sequence with cmpld/setb for power9.
39468 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
39469 for generating subfc. instruction.
39470 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
39471 now uses this instruction.
39473 2017-01-30 Ian Lance Taylor <iant@google.com>
39476 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
39477 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
39479 2017-01-30 Martin Sebor <msebor@redhat.com>
39481 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
39482 Move constant to the right of a relational operator.
39483 (get_mpfr_format_length, format_character, format_string): Ditto.
39484 (should_warn_p, maybe_warn): Same.
39486 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
39488 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
39491 * asan.c (get_translation_unit_decl): Remove function.
39492 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
39494 2017-01-30 Martin Liska <mliska@suse.cz>
39496 PR gcov-profile/79259
39497 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
39498 -fprofile-generate.
39500 2017-01-30 Martin Liska <mliska@suse.cz>
39503 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
39504 Initialize variables with NULL value.
39506 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
39509 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
39511 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
39513 2017-01-30 Richard Biener <rguenther@suse.de>
39515 PR tree-optimization/79276
39516 * tree-vrp.c (process_assert_insertions): Properly adjust common
39517 when removing a duplicate.
39519 2017-01-30 Richard Biener <rguenther@suse.de>
39521 PR tree-optimization/79256
39522 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
39523 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
39525 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
39527 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
39530 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
39531 ("*r<noxa>sbg_<mode>_sll_bitmask")
39532 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
39533 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
39534 Use contiguous_bitmask_nowrap_operand.
39536 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
39539 * config/rs6000/altivec.h (vec_xl): Revise #define.
39540 (vec_xst): Likewise.
39542 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
39544 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
39546 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
39548 PR rtl-optimization/79194
39549 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
39550 traps before call to bypass_conditional_jumps.
39552 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39554 PR tree-optimization/71374
39555 * lra-constraints.c (check_conflict_input_operands): New.
39556 (match_reload): Use it.
39558 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
39561 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
39562 account to calculate conflict_set.
39564 2017-01-27 Bin Cheng <bin.cheng@arm.com>
39566 PR rtl-optimization/78559
39567 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
39568 other_insn in combine.
39570 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
39572 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
39573 uint16_type_node for BT_UINT16.
39575 2017-01-27 David Malcolm <dmalcolm@redhat.com>
39577 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
39578 "RTL Tests" to menu.
39579 (GIMPLE Tests): New node.
39580 (RTL Tests): New node.
39582 2017-01-27 Richard Biener <rguenther@suse.de>
39584 PR tree-optimization/79245
39585 * tree-loop-distribution.c (distribute_loop): Apply cost
39586 modeling also to detected patterns.
39588 2017-01-27 Richard Biener <rguenther@suse.de>
39590 PR tree-optimization/71433
39591 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
39592 (compare_assert_loc): New function.
39593 (process_assert_insertions): Sort and optimize assert locations
39594 to remove duplicates and push down identical assertions on
39595 edges to their destination block.
39597 2017-01-27 Richard Biener <rguenther@suse.de>
39599 PR tree-optimization/79244
39600 * tree-vrp.c (remove_range_assertions): Forcefully propagate
39601 out SSA names even if abnormal.
39603 2017-01-27 Jakub Jelinek <jakub@redhat.com>
39605 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
39606 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
39607 instead of MPFR_RNDN.
39609 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
39612 * arm.c (arm_option_override): Don't call build_target_option_node
39613 until after doing all option overrides.
39614 (arm_valid_target_attribute_tree): Likewise.
39616 2017-01-27 Martin Liska <mliska@suse.cz>
39618 * doc/invoke.texi (-fprofile-arcs): Document profiling support
39619 for {cd}tors and C++ {cd}tors.
39621 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
39623 * config/s390/s390.md ("*setmem_long_and")
39624 ("*setmem_long_and_31z"): Use zero_extend instead of and.
39626 2017-01-26 Martin Sebor <msebor@redhat.com>
39628 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
39631 2017-01-26 Martin Sebor <msebor@redhat.com>
39633 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
39634 HAVE_DFmode before using XFmode or DFmode.
39635 (parse_directive): Avoid using the z length modifier to avoid
39636 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
39638 PR middle-end/78703
39639 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
39640 to accept adjustment as an array.
39641 (get_int_range): New function.
39642 (struct directive): Make width and prec arrays.
39643 (directive::set_width, directive::set_precision): Call get_int_range.
39644 (format_integer, format_floating): Handle width and precision ranges.
39645 (format_string, parse_directive): Same.
39647 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39650 * dwarf2out.c (generate_skeleton_bottom_up): For children with
39651 comdat_type_p set, just clone them, but keep the children in the
39655 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
39656 which have direct callers with -fvar-tracking-assignments enabled
39658 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
39659 inside of type units.
39661 2017-01-26 Martin Sebor <msebor@redhat.com>
39663 PR middle-end/78703
39664 * gimple-ssa-sprintf.c (struct result_range): Add likely and
39666 (struct format_result): Replace number_chars, number_chars_min,
39667 and number_chars_max with a single member of struct result_range.
39669 (format_result::operator+=): Adjust.
39670 (struct fmtresult): Remove bounded. Handle likely and unlikely
39672 (fmtresult::adjust_for_width_or_precision): New function.
39673 (fmtresult:type_max_digits): New function.
39674 (bytes_remaining): Handle likely and unlikely counters.
39675 (min_bytes_remaining): Remove.
39676 (format_percent): Simplify.
39677 (format_integer, format_floating): Set likely and unlikely counters.
39678 (get_string_length, format_character, format_string): Same.
39679 (format_plain, should_warn_p): New function.
39680 (maybe_warn): Call should_warn_p. Update diagnostic messages
39681 and handle those for all directives, including plain strings.
39682 (format_directive): Handle likely and unlikely counters.
39683 Remove unnecessary quoting from diagnostics. Add an informational
39685 (add_bytes): Remove.
39686 (pass_sprintf_length::compute_format_length): Simplify.
39687 (try_substitute_return_value): Handle likely and unlikely counters.
39689 2017-01-26 Carl Love <cel@us.ibm.com>
39691 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
39692 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
39694 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
39697 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
39698 endianess for subregs into account.
39699 * lra-constraints.c (lra_constraints): Do risky transformations
39700 always on the first iteration.
39701 * lra-lives.c (check_pseudos_live_through_calls): Add arg
39702 last_call_used_reg_set.
39703 (process_bb_lives): Define and use last_call_used_reg_set.
39704 * lra.c (lra): Always continue after lra_constraints on the first
39707 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
39709 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
39711 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
39713 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39715 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
39716 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
39717 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
39718 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
39719 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
39720 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
39721 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
39722 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
39723 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
39725 2017-01-26 Marek Polacek <polacek@redhat.com>
39728 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
39729 for the third operand.
39731 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39733 PR middle-end/79236
39734 * omp-low.c (struct omp_context): Add simt_stmt field.
39735 (scan_omp_for): Return omp_context *.
39736 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
39737 context to the _simt_ SIMD stmt.
39738 (lower_omp_for): For combined SIMD with sibling _simt_
39739 SIMD, make sure to use the same decls in _looptemp_
39740 clauses as in the sibling.
39742 2017-01-26 David Sherwood <david.sherwood@arm.com>
39744 PR middle-end/79212
39745 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
39748 2017-01-26 Jakub Jelinek <jakub@redhat.com>
39751 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
39752 emit fld b; fld a; if possible.
39754 * brig-builtins.def: Update copyright years.
39755 * config/arm/arm_acle_builtins.def: Update copyright years.
39757 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
39760 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
39761 constraint instead of o for the stxsd instruction.
39763 2017-01-25 Carl Love <cel@us.ibm.com>
39765 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
39766 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
39768 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
39770 * doc/invoke.texi (C++ Dialect Options): Fix typo.
39772 2017-01-25 Richard Biener <rguenther@suse.de>
39774 PR tree-optimization/69264
39775 * target.def (vector_alignment_reachable): Improve documentation.
39776 * doc/tm.texi: Regenerate.
39777 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
39779 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
39780 earlier changes with respect to TYPE_USER_ALIGN.
39781 (vector_alignment_reachable_p): Likewise. Improve dumping.
39783 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39786 * config/arm/arm.md (xordi3): Force constant operand into a register
39789 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
39791 * doc/invoke.texi (-fstore-merging): Correct default optimization
39792 levels at which it is enabled.
39793 (-O): Move -fstore-merging from list to...
39796 2017-01-25 Richard Biener <rguenther@suse.de>
39799 * omp-expand.c: Include debug.h.
39800 (expand_omp_taskreg): Make sure to generate early debug before
39801 outlining anything from a function.
39802 (expand_omp_target): Likewise.
39803 (grid_expand_target_grid_body): Likewise.
39805 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
39808 * asan.c (get_translation_unit_decl): New function.
39809 (asan_add_global): Extract modules file name from globals
39810 TRANSLATION_UNIT_DECL name.
39812 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
39815 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
39816 for long calls with APCS frame and VFP.
39818 2017-01-24 David Malcolm <dmalcolm@redhat.com>
39820 * cfg.c (original_copy_tables_initialized_p): New function.
39821 * cfg.h (original_copy_tables_initialized_p): New decl.
39822 * cfgrtl.c (relink_block_chain): Guard the call to
39823 free_original_copy_tables with a call to
39824 original_copy_tables_initialized_p.
39825 * cgraph.h (symtab_node::native_rtl_p): New decl.
39826 * cgraphunit.c (symtab_node::native_rtl_p): New function.
39827 (symtab_node::needed_p): Don't assert for early assembly output
39828 for __RTL functions.
39829 (cgraph_node::finalize_function): Set "force_output" for __RTL
39831 (cgraph_node::analyze): Bail out early for __RTL functions.
39832 (analyze_functions): Update assertion to support __RTL functions.
39833 (cgraph_node::expand): Bail out early for __RTL functions.
39834 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
39836 * function.h (struct function): Update comment for field
39838 * gimple-expr.c: Include "tree-pass.h".
39839 (gimple_has_body_p): Return false for __RTL functions.
39840 * Makefile.in (OBJS): Add run-rtl-passes.o.
39841 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
39843 (gcc::pass_manager::get_clean_slate): New accessor.
39844 * passes.c: Include "insn-addr.h".
39845 (should_skip_pass_p): Add logging. Update logic for running
39846 "expand" to be compatible with both __GIMPLE and __RTL. Guard
39847 property-provider override so it is only done for gimple passes.
39849 (skip_pass): New function.
39850 (execute_one_pass): Call skip_pass when skipping passes.
39851 * read-md.c (md_reader::read_char): Support filtering
39852 the input to a subset of line numbers.
39853 (md_reader::md_reader): Initialize fields
39854 m_first_line and m_last_line.
39855 (md_reader::read_file_fragment): New function.
39856 * read-md.h (md_reader::read_file_fragment): New decl.
39857 (md_reader::m_first_line): New field.
39858 (md_reader::m_last_line): New field.
39859 * read-rtl-function.c (function_reader::create_function): Only
39860 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
39861 curr_properties. Set DECL_INITIAL to a dummy block.
39862 (read_rtl_function_body_from_file_range): New function.
39863 * read-rtl-function.h (read_rtl_function_body_from_file_range):
39865 * run-rtl-passes.c: New file.
39866 * run-rtl-passes.h: New file.
39868 2017-01-24 Jeff Law <law@redhat.com>
39870 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
39873 2017-01-24 Bin Cheng <bin.cheng@arm.com>
39875 PR tree-optimization/79159
39876 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
39877 (record_nonwrapping_iv): Improve boundary using above function if no
39878 value range information.
39880 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
39881 Martin Jambor <mjambor@suse.cz>
39883 * brig-builtins.def: New file.
39884 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
39885 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
39886 (DEF_HSAIL_SAT_BUILTIN): Likewise.
39887 (DEF_HSAIL_INTR_BUILTIN): Likewise.
39888 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
39889 * builtin-types.def (BT_INT8): New.
39890 (BT_INT16): Likewise.
39891 (BT_UINT8): Likewise.
39892 (BT_UINT16): Likewise.
39893 (BT_FN_ULONG): Likewise.
39894 (BT_FN_UINT_INT): Likewise.
39895 (BT_FN_UINT_ULONG): Likewise.
39896 (BT_FN_UINT_LONG): Likewise.
39897 (BT_FN_UINT_PTR): Likewise.
39898 (BT_FN_ULONG_PTR): Likewise.
39899 (BT_FN_INT8_FLOAT): Likewise.
39900 (BT_FN_INT16_FLOAT): Likewise.
39901 (BT_FN_UINT32_FLOAT): Likewise.
39902 (BT_FN_UINT16_FLOAT): Likewise.
39903 (BT_FN_UINT8_FLOAT): Likewise.
39904 (BT_FN_UINT64_FLOAT): Likewise.
39905 (BT_FN_UINT16_UINT32): Likewise.
39906 (BT_FN_UINT32_UINT16): Likewise.
39907 (BT_FN_UINT16_UINT16_UINT16): Likewise.
39908 (BT_FN_INT_PTR_INT): Likewise.
39909 (BT_FN_UINT_PTR_UINT): Likewise.
39910 (BT_FN_LONG_PTR_LONG): Likewise.
39911 (BT_FN_ULONG_PTR_ULONG): Likewise.
39912 (BT_FN_VOID_UINT64_UINT64): Likewise.
39913 (BT_FN_UINT8_UINT8_UINT8): Likewise.
39914 (BT_FN_INT8_INT8_INT8): Likewise.
39915 (BT_FN_INT16_INT16_INT16): Likewise.
39916 (BT_FN_INT_INT_INT): Likewise.
39917 (BT_FN_UINT_FLOAT_UINT): Likewise.
39918 (BT_FN_FLOAT_UINT_UINT): Likewise.
39919 (BT_FN_ULONG_UINT_UINT): Likewise.
39920 (BT_FN_ULONG_UINT_PTR): Likewise.
39921 (BT_FN_ULONG_ULONG_ULONG): Likewise.
39922 (BT_FN_UINT_UINT_UINT): Likewise.
39923 (BT_FN_VOID_UINT_PTR): Likewise.
39924 (BT_FN_UINT_UINT_PTR: Likewise.
39925 (BT_FN_UINT32_UINT64_PTR): Likewise.
39926 (BT_FN_INT_INT_UINT_UINT): Likewise.
39927 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
39928 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
39929 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
39930 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
39931 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
39932 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
39933 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
39934 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
39935 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
39936 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
39937 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
39938 * doc/frontends.texi: List BRIG FE.
39939 * doc/install.texi (Testing): Add BRIG tesring requirements.
39940 * doc/invoke.texi (Overall Options): Mention BRIG.
39941 * doc/standards.texi (Standards): Doucment BRIG HSA version.
39943 2017-01-24 Richard Biener <rguenther@suse.de>
39945 PR translation/79208
39946 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
39948 2017-01-24 Martin Jambor <mjambor@suse.cz>
39951 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
39952 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
39953 and known_contexts.
39955 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
39957 PR middle-end/79123
39958 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
39959 casts from signed to unsigned really don't have a range.
39961 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
39963 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
39964 GMP_RNDx for compatiblity.
39966 2017-01-24 Martin Liska <mliska@suse.cz>
39969 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
39970 that would prevent us to call alloca with -1 as argument.
39972 2017-01-24 Jakub Jelinek <jakub@redhat.com>
39974 * dwarf2out.c (output_compilation_unit_header, output_file_names):
39975 Avoid -Wformat-security warning.
39977 2017-01-23 Andrew Pinski <apinski@cavium.com>
39979 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
39982 2017-01-23 Martin Sebor <msebor@redhat.com>
39984 PR middle-end/78703
39985 * gimple-ssa-sprintf.c (warn_level): New global.
39986 (format_integer): Use it here and throughout the rest of the file.
39987 Use the same switch to compute sign as base.
39988 (maybe_warn): New function.
39989 (format_directive): Factor out warnings into maybe_warn.
39990 Add debugging output. Use warn_level.
39991 (add_bytes): Use warn_level.
39992 (pass_sprintf_length::compute_format_length): Add debugging output.
39993 (try_substitute_return_value): Same.
39994 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
39996 PR middle-end/78703
39997 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
39998 (struct fmtresult, format_integer, format_floating): Adjust.
39999 (fmtresult::fmtresult): Set max correctly in two argument ctor.
40000 (get_string_length, format_string,format_directive): Same.
40001 (pass_sprintf_length::compute_format_length): Same.
40002 (try_substitute_return_value): Simplify slightly.
40004 PR middle-end/78703
40005 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
40006 (fmtresult::operator+=): Outlined.
40007 (struct fmtresult): Add ctors.
40008 (struct conversion_spec): Rename...
40009 (struct directive): ...to this. Add and remove data members.
40010 (directive::set_width, directive::set_precision): New functions.
40011 (format_percent): Use fmtresult ctor.
40012 (get_width_and_precision): Remove.
40013 (format_integer): Make naming changes. Avoid computing width and
40015 (format_floating): Same. Adjust indentation.
40016 (format_character, format_none): New functions.
40017 (format_string): Moved character handling to format_character.
40018 (format_directive): Remove arguments, change return type.
40019 (parse_directive): New function.
40020 (pass_sprintf_length::compute_format_length): Move directive
40021 parsing to parse_directive.
40023 2017-01-23 Jakub Jelinek <jakub@redhat.com>
40025 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
40026 (assign_assembler_name_if_needed): ... this.
40027 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
40028 (assign_assembler_name_if_needed): ... this.
40029 (free_lang_data_in_cgraph): Adjust callers.
40030 * cgraphunit.c (cgraph_node::analyze): Likewise.
40031 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
40034 2017-01-23 Richard Biener <rguenther@suse.de>
40036 PR tree-optimization/79088
40037 PR tree-optimization/79188
40038 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
40039 resetting loop bounds after last path deletion. Reset loop
40040 bounds of the target loop, make code match the comments.
40041 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
40042 Make sure loops need no fixups.
40044 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
40046 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
40047 exponent support with double type for first argument.
40048 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
40049 type returned by __builtin_vec_extract_sig,
40050 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
40051 functions from "vector int" to "vector unsigned int" or from
40052 "vector long long int" to "vector unsigned long long int".
40053 Changed type returned by __builtin_vec_extract_exp,
40054 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
40055 functions from "vector int" to "vector unsigned int" or from
40056 "vector long long int" to "vector unsigned long long int".
40057 Changed return type of __builtin_vec_test_data_class,
40058 __builtin_vec_test_data_class_sp, and
40059 __builtin_vec_test_data_class_dp from "vector int" to
40060 "vector bool int" or from "vector long long int" to "vector bool
40061 long long int" and changed second argument type from "unsigned
40062 int" to "int". Added new overloaded function forms "vector float
40063 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
40064 "vector float __builtin_vec_insert_exp_sp (vector float, vector
40065 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
40066 double, vector unsigned long long int)" and "vector double
40067 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
40068 long int)". Changed return type of
40069 __builtin_scalar_test_data_class and
40070 __builtin_scalar_test_data_class_sp and
40071 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
40072 int" and changed second argument from "unsigned int" to "int".
40073 Changed type returned by __builtin_scalar_test_neg,
40074 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
40075 from "int" to "bool int". Added new overloaded function form
40076 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
40077 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
40078 exponent double-precision with floating point first argument.
40079 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
40080 documentation of scalar_test_data_class, scalar_test_neg,
40081 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
40082 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
40083 vec_test_data_class built-in functions to reflect refinements in
40084 their type signatures.
40086 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
40088 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
40090 (aarch64_elf_asm_destructor): Likewise.
40092 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
40094 PR rtl-optimization/78634
40095 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
40096 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
40097 * ifcvt.c (noce_try_cmove): Add missing cost check.
40099 PR rtl-optimization/71724
40100 * combine.c (if_then_else_cond): Look for situations where it is
40101 beneficial to undo the work of one of the recursive calls.
40103 2017-01-23 Bin Cheng <bin.cheng@arm.com>
40105 PR tree-optimization/70754
40106 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
40107 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
40108 combined stmt before it if not NULL.
40109 (combine_chains): Process refs reversely and compute dominance point
40112 2017-01-23 Martin Liska <mliska@suse.cz>
40114 PR tree-optimization/79196
40115 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
40116 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
40118 (strlen_optimize_stmt): Call the renamed function.
40120 2017-01-23 Michael Matz <matz@suse.de>
40122 PR tree-optimization/78384
40123 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
40125 2017-01-23 Richard Biener <rguenther@suse.de>
40127 PR tree-optimization/79186
40128 * tree-vrp.c (register_new_assert_for): Make sure we've seen
40129 both incoming edges before moving an assert.
40131 2017-01-23 Martin Jambor <mjambor@suse.cz>
40133 * ipa-prop.c (load_from_param_1): Removed.
40134 (load_from_unmodified_param): Bits from load_from_param_1 put back
40136 (load_from_param): Removed.
40137 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
40138 with stmt. Reverted back to use of load_from_unmodified_param.
40140 2017-01-23 Martin Jambor <mjambor@suse.cz>
40143 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
40144 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
40145 field a pointer to garbage collected vector, mark lattices and
40146 ipcp_orig_node with GTY((skip)).
40147 (ipa_get_param_count): Adjust to descriptors being a pointer.
40148 (ipa_get_param): Likewise.
40149 (ipa_get_type): Likewise.
40150 (ipa_get_param_move_cost): Likewise.
40151 (ipa_set_param_used): Likewise.
40152 (ipa_get_controlled_uses): Likewise.
40153 (ipa_set_controlled_uses): Likewise.
40154 (ipa_is_param_used): Likewise.
40155 (ipa_node_params_t): Move into garbage collector. New methods insert
40157 (ipa_node_params_sum): Annotate wth GTY(()).
40158 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
40160 (ipa_load_from_parm_agg): Adjust declaration.
40161 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
40162 * ipa-profile.c (ipa_profile): Likewise.
40163 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
40164 (ipa_populate_param_decls): Make descriptors parameter garbage
40166 (ipa_dump_param): Adjust to descriptors being a pointer.
40167 (ipa_alloc_node_params): Likewise.
40168 (ipa_initialize_node_params): Likewise.
40169 (load_from_param_1): Make descriptors parameter garbage collected.
40170 (load_from_unmodified_param): Likewise.
40171 (load_from_param): Likewise.
40172 (ipa_load_from_parm_agg): Likewise.
40173 (ipa_node_params::~ipa_node_params): Removed.
40174 (ipa_free_all_node_params): Remove call to delete operator.
40175 (ipa_node_params_t::insert): New.
40176 (ipa_node_params_t::remove): Likewise.
40177 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
40178 copy known_csts and known_contexts vectors.
40179 (ipa_read_node_info): Adjust to descriptors being a pointer.
40180 (ipcp_modif_dom_walker): Make m_descriptors field garbage
40182 (ipcp_transform_function): Make descriptors variable garbage
40185 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
40187 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
40188 * config/i386/avx512dqintrin.h: Ditto.
40189 * config/i386/avx512fintrin.h: Ditto.
40190 * config/i386/i386.c: Handle new builtins.
40191 * config/i386/i386-builtin.def: Add new builtins.
40192 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
40193 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
40195 2017-01-23 Jakub Jelinek <jakub@redhat.com>
40196 Martin Liska <mliska@suse.cz>
40198 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
40199 * asan.c (asan_expand_poison_ifn): Support stores and use
40200 appropriate ASAN report function.
40201 * internal-fn.c (expand_ASAN_POISON_USE): New function.
40202 * internal-fn.def (ASAN_POISON_USE): Declare.
40203 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
40204 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
40205 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
40206 ASAN_POISON calls w/o LHS.
40207 * tree-ssa.c (execute_update_addresses_taken): Create clobber
40208 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
40209 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
40210 * gimplify.c (asan_poison_variables): Add attribute
40211 use_after_scope_memory to variables that really needs to live
40213 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
40214 having the attribute.
40216 2017-01-23 Martin Liska <mliska@suse.cz>
40218 * asan.c (create_asan_shadow_var): New function.
40219 (asan_expand_poison_ifn): Likewise.
40220 * asan.h (asan_expand_poison_ifn): New declaration.
40221 * internal-fn.c (expand_ASAN_POISON): Likewise.
40222 * internal-fn.def (ASAN_POISON): New builtin.
40223 * sanopt.c (pass_sanopt::execute): Expand
40224 asan_expand_poison_ifn.
40225 * tree-inline.c (copy_decl_for_dup_finish): Make function
40227 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
40228 * tree-ssa.c (is_asan_mark_p): New function.
40229 (execute_update_addresses_taken): Rewrite local variables
40230 (identified just by use-after-scope as addressable) into SSA.
40232 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40234 * doc/install.texi (Specific): opensource.apple.com uses https
40235 now. Remove trailing slash.
40237 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
40239 * README.Portability: Remove note on an Irix compatibility issue.
40241 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
40243 * gcov.c (INCLUDE_ALGORITHM): Define.
40244 (INCLUDE_VECTOR): Define.
40245 No longer include <vector> and <algorithm> directly.
40247 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
40249 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
40251 * doc/invoke.texi (Code Gen Options): Ditto.
40253 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
40256 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
40258 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
40260 rtl-optimization/79125
40261 * cprop.c (local_cprop_pass): Handle cases where we make an
40262 unconditional trap.
40264 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
40268 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
40269 read from, for big endian.
40271 2017-01-20 Jiong Wang <jiong.wang@arm.com>
40273 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
40274 register pauth builtins for LP64 only.
40276 2017-01-20 Marek Polacek <polacek@redhat.com>
40279 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
40282 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40284 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
40286 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
40287 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
40288 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
40290 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40293 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
40294 in big-endian mode when they are not a single duplicated value.
40296 2017-01-20 Richard Biener <rguenther@suse.de>
40298 * BASE-VER: Bump to 7.0.1.
40300 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
40302 * omp-low.c (omplow_simd_context): New struct. Use it...
40303 (lower_rec_simd_input_clauses): ...here and...
40304 (lower_rec_input_clauses): ...here to hold common data. Adjust all
40305 references to idx, lane, max_vf, is_simt.
40307 2017-01-20 Graham Markall <graham.markall@embecosm.com>
40309 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
40312 2017-01-20 Martin Jambor <mjambor@suse.cz>
40314 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
40315 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
40317 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
40318 (GTFILES): Rename hsa.c to hsa-common.c.
40319 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
40320 * hsa-dump.c: Likewise.
40321 * hsa-gen.c: Likewise.
40322 * hsa-regalloc.c: Likewise.
40323 * ipa-hsa.c: Likewise.
40324 * omp-expand.c: Likewise.
40325 * omp-low.c: Likewise.
40326 * toplev.c: Likewise.
40328 2017-01-20 Marek Polacek <polacek@redhat.com>
40331 * doc/invoke.texi: Document -Wduplicated-branches.
40332 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
40333 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
40334 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
40335 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
40336 return 0 only when not OEP_LEXICOGRAPHIC.
40337 (fold_build_cleanup_point_expr): Use the expression
40338 location when building CLEANUP_POINT_EXPR.
40339 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
40340 * tree.c (add_expr): Handle error_mark_node.
40342 2017-01-20 Martin Liska <mliska@suse.cz>
40345 * tree-profile.c (init_ic_make_global_vars): Do not call
40347 (gimple_init_gcov_profiler): Likewise.
40349 2017-01-20 Martin Liska <mliska@suse.cz>
40352 * cgraph.h (maybe_create_reference): Remove argument and
40354 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
40356 * ipa-cp.c (create_specialized_node): Likewise.
40357 * symtab.c (symtab_node::maybe_create_reference): Handle
40358 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
40360 2017-01-20 Martin Liska <mliska@suse.cz>
40362 * read-rtl-function.c (function_reader::create_function): Use
40363 build_decl instread of build_decl_stat.
40365 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
40367 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
40368 * config/i386/avx512dqintrin.h: Ditto.
40369 * config/i386/avx512fintrin.h: Ditto.
40370 * config/i386/i386-builtin-types.def: Add new types.
40371 * config/i386/i386.c: Handle new types.
40372 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
40373 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
40374 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
40375 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
40376 (__builtin_ia32_kshiftridi): New.
40377 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
40379 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
40383 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
40384 define to rs6000_init_stack_protect_guard.
40385 (rs6000_init_stack_protect_guard): New function.
40387 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40388 Yunqiang Su <yunqiang.su@imgtec.com>
40390 * config.gcc (supported_defaults): Add madd4.
40391 (with_madd4): Add validation.
40392 (all_defaults): Add madd4.
40393 * config/mips/mips.opt (mmadd4): New option.
40394 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40396 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40398 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
40399 (ISA_HAS_FUSED_MADD4): Likewise.
40400 * doc/invoke.texi (-mmadd4): Document the new option.
40401 * doc/install.texi (--with-madd4): Document the new option.
40403 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40405 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
40406 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
40407 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
40408 (aarch64_init_pauth_hint_builtins): New.
40409 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
40410 (aarch64_expand_builtin): Expand new builtins.
40412 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40414 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
40415 * combine-stack-adj.c (no_unhandled_cfa): Handle
40416 REG_CFA_TOGGLE_RA_MANGLE.
40417 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
40418 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
40419 info for return address signing.
40420 (aarch64_expand_epilogue): Likewise.
40422 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40424 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
40425 * config/aarch64/aarch64-protos.h
40426 (aarch64_return_address_signing_enabled): New declaration.
40427 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
40429 (aarch64_expand_prologue): Sign return address before it's pushed onto
40431 (aarch64_expand_epilogue): Authenticate return address fetched from
40433 (aarch64_override_options): Sanity check for ILP32 and ISA level.
40434 (aarch64_attributes): New function attributes for "sign-return-address".
40435 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
40436 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
40437 ("*do_return"): Generate combined instructions according to key index.
40438 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
40439 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
40441 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
40442 * config/aarch64/aarch64.opt (msign-return-address=): New.
40443 * doc/extend.texi (AArch64 Function Attributes): Documents
40444 "sign-return-address=".
40445 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
40447 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
40449 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
40450 overall option summary.
40452 2017-01-19 Jiong Wang <jiong.wang@arm.com>
40454 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
40455 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
40456 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
40457 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
40459 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
40461 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
40462 -mpower9-minmax by default for -mcpu=power9.
40463 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
40464 128-bit floating point.
40466 2017-01-20 Alan Modra <amodra@gmail.com>
40468 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
40469 optimizing for size.
40471 2017-01-20 Alan Modra <amodra@gmail.com>
40474 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
40475 for strcmp and strncmp from corresponding builtin decl.
40477 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40479 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
40480 instead of i386/rtems-64.h.
40481 * config/i386/rtems-64.h: Remove.
40483 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
40487 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
40489 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
40491 2017-01-19 Tamar Christina <tamar.christina@arm.com>
40493 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
40494 Change int to HOST_WIDE_INT.
40495 * config/aarch64/aarch64-protos.h
40496 (aarch64_simd_gen_const_vector_dup): Likewise.
40497 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
40499 2017-01-19 David Malcolm <dmalcolm@redhat.com>
40501 * langhooks-def.h (lhd_type_for_size): New decl.
40502 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
40503 * langhooks.c (lhd_type_for_size): New function, taken from
40506 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
40508 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
40509 define_bypass for CR latency.
40510 (power9-cracked-alu): Update bypass latency and remove power9-branch.
40511 (power9-alu2): Add define_bypass for CR latency.
40513 (power9-mul): Update insn latency.
40514 (power9-mul-compare): Update insn latency, bypass latency and remove
40517 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
40519 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
40521 * config/aarch64/aarch64.md
40522 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
40523 aarch64_nopcrelative_literal_loads.
40524 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
40526 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
40528 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
40529 TARGET_LOONGSON_3A.
40530 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
40532 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
40535 * config.gcc (supported_defaults): Add lxc1-sxc1.
40536 (with_lxc1_sxc1): Add validation.
40537 (all_defaults): Add lxc1-sxc1.
40538 * config/mips/mips.opt (mlxc1-sxc1): New option.
40539 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
40541 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
40542 __mips_no_lxc1_sxc1.
40543 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
40544 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
40545 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
40547 2017-01-19 Richard Biener <rguenther@suse.de>
40549 PR tree-optimization/72488
40550 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
40551 sure to restore SSA info.
40552 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
40554 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
40556 PR rtl-optimization/79121
40557 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
40558 of the inner type when shifting an extended value.
40560 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40563 * symtab.c (symtab_node::equal_address_to): Fix comparing of
40564 interposable aliases.
40566 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
40569 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
40570 Use the evmergelohi instruction.
40571 (mov_si<mode>_e500_subreg4_2_le): Likewise.
40572 (mov_sitf_e500_subreg8_2_be): Likewise.
40573 (mov_sitf_e500_subreg12_2_le): Likewise.
40574 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
40575 (mov_si<mode>_e500_subreg4_2_be): Likewise.
40576 (mov_sitf_e500_subreg8_2_le): Likewise.
40577 (mov_sitf_e500_subreg12_2_be): Likewise.
40579 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40581 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
40582 attribute from vecsimple to vecperm.
40583 (altivec_vbpermq2): Likewise.
40585 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40588 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
40590 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
40591 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
40592 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
40593 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
40594 case where N arg is SIZE_MAX.
40595 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
40596 (cmpstrsi): Add pattern.
40598 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
40600 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
40601 __builtin_vec_revb builtins.
40602 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
40603 built-in functions to support generation of the ISA 3.0 XXBR<x>
40604 vector byte reverse instructions.
40605 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
40606 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
40607 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
40608 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
40609 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
40610 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
40611 (P9V_BUILTIN_VEC_REVB): Likewise.
40612 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
40613 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
40614 (p9_xxbrq_v16qi): Likewise.
40615 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
40616 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
40617 (p9_xxbrh_v8hi): Likewise.
40618 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
40619 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
40620 vec_revb built-in functions.
40622 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
40624 PR rtl-optimization/78952
40625 * config/i386/i386.md (any_extract): New code iterator.
40626 (*insvqi_2): Use any_extract for source operand.
40627 (*insvqi_3): Use any_shiftrt for source operand.
40629 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
40631 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
40633 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
40635 2017-01-18 Matthias Klose <doko@ubuntu.com>
40637 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
40639 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40641 * config/rs6000/altivec.h (vec_bperm): Change #define.
40642 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
40643 (altivec_vbpermq2): New define_insn.
40644 (altivec_vbpermd): Likewise.
40645 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
40646 function interface.
40647 (VBPERMD): Likewise.
40648 (VBPERM): New polymorphic function interface.
40649 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
40650 Add entries for P9V_BUILTIN_VEC_VBPERM.
40651 * doc/extend.texi: Add interfaces for vec_bperm.
40653 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
40655 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
40656 first letter of error messages.
40657 (s390_resolve_overloaded_builtin): Likewise.
40658 * config/s390/s390.c (s390_expand_builtin): Likewise.
40659 (s390_invalid_arg_for_unprototyped_fn): Likewise.
40660 (s390_valid_target_attribute_inner_p): Likewise.
40661 * config/s390/s390.md ("tabort"): Likewise.
40663 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
40665 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
40666 (ISA_AVOID_DIV_HILO): New macro.
40667 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
40668 (ISA_HAS_DDIV): Likewise.
40670 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40672 * doc/invoke.texi (fabi-version): Correct number of occurrences.
40674 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40676 * doc/invoke.texi (fabi-version): Spelling fix.
40678 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40681 * doc/invoke.texi (fabi-version): Mention mangling fix for
40684 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
40687 * doc/invoke.texi (fabi-version): Document discriminator mangling.
40689 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
40692 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
40693 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
40695 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
40697 (stack_protect_test): Ditto.
40698 * config/rs6000/rs6000.opt (mstack-protector-guard=,
40699 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
40701 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
40702 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
40703 -mstack-protector-guard-offset=.
40704 (RS/6000 and PowerPC Options): Ditto.
40706 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
40708 * config/i386/i386.h (MASK_CLASS_P): New define.
40709 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
40710 there are no registers from different register sets also when
40711 mask registers are used. Update function comment.
40712 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
40713 to (*k/*r) and (*k/*km) alternatives.
40715 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
40717 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
40718 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
40719 (EH_RETURN_HANDLER_RTX): New define.
40720 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
40721 Force frame pointer in EH return functions.
40722 (aarch64_expand_epilogue): Add barrier for eh_return.
40723 (aarch64_final_eh_return_addr): Remove.
40724 (aarch64_eh_return_handler_rtx): New function.
40725 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
40727 (aarch64_eh_return_handler_rtx): New prototype.
40729 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
40731 * config/rs6000/altivec.h (vec_rlmi): New #define.
40732 (vec_vrlnm): Likewise.
40733 (vec_rlnm): Likewise.
40734 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
40735 (UNSPEC_VRLNM): Likewise.
40736 (VIlong): New mode iterator.
40737 (altivec_vrl<VI_char>mi): New define_insn.
40738 (altivec_vrl<VI_char>nm): Likewise.
40739 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
40741 (VRLDNM): Likewise.
40742 (RLNM): New polymorphic function entry.
40743 (VRLWMI): New monomorphic function entry.
40744 (VRLDMI): Likewise.
40745 (RLMI): New polymorphic function entry.
40746 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
40747 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
40748 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
40751 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40754 * dwarf2out.c (field_byte_offset): Restore the
40755 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
40756 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
40757 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
40760 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40763 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
40765 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40768 * dwarf2out.c (add_data_member_location_attribute): For constant
40769 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
40770 instead of DW_AT_data_member_location, DW_AT_bit_offset and
40771 DW_AT_byte_size attributes.
40773 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
40775 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
40776 after forcing to constant memory when the code model is medium.
40778 2017-01-17 Julia Koval <julia.koval@intel.com>
40781 * config/i386/avx512fintrin.h
40782 (_mm512_i32gather_ps): Change __addr type to void const*.
40783 (_mm512_mask_i32gather_ps): Ditto.
40784 (_mm512_i32gather_pd): Ditto.
40785 (_mm512_mask_i32gather_pd): Ditto.
40786 (_mm512_i64gather_ps): Ditto.
40787 (_mm512_mask_i64gather_ps): Ditto.
40788 (_mm512_i64gather_pd): Ditto.
40789 (_mm512_mask_i64gather_pd): Ditto.
40790 (_mm512_i32gather_epi32): Ditto.
40791 (_mm512_mask_i32gather_epi32): Ditto.
40792 (_mm512_i32gather_epi64): Ditto.
40793 (_mm512_mask_i32gather_epi64): Ditto.
40794 (_mm512_i64gather_epi32): Ditto.
40795 (_mm512_mask_i64gather_epi32): Ditto.
40796 (_mm512_i64gather_epi64): Ditto.
40797 (_mm512_mask_i64gather_epi64): Ditto.
40798 (_mm512_i32scatter_ps): Change __addr type to void*.
40799 (_mm512_mask_i32scatter_ps): Ditto.
40800 (_mm512_i32scatter_pd): Ditto.
40801 (_mm512_mask_i32scatter_pd): Ditto.
40802 (_mm512_i64scatter_ps): Ditto.
40803 (_mm512_mask_i64scatter_ps): Ditto.
40804 (_mm512_i64scatter_pd): Ditto.
40805 (_mm512_mask_i64scatter_pd): Ditto.
40806 (_mm512_i32scatter_epi32): Ditto.
40807 (_mm512_mask_i32scatter_epi32): Ditto.
40808 (_mm512_i32scatter_epi64): Ditto.
40809 (_mm512_mask_i32scatter_epi64): Ditto.
40810 (_mm512_i64scatter_epi32): Ditto.
40811 (_mm512_mask_i64scatter_epi32): Ditto.
40812 (_mm512_i64scatter_epi64): Ditto.
40813 (_mm512_mask_i64scatter_epi64): Ditto.
40814 * config/i386/avx512pfintrin.h
40815 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
40816 (_mm512_mask_prefetch_i32gather_ps): Ditto.
40817 (_mm512_mask_prefetch_i64gather_pd): Ditto.
40818 (_mm512_mask_prefetch_i64gather_ps): Ditto.
40819 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
40820 (_mm512_prefetch_i32scatter_ps): Ditto.
40821 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
40822 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
40823 (_mm512_prefetch_i64scatter_pd): Ditto.
40824 (_mm512_prefetch_i64scatter_ps): Ditto.
40825 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
40826 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
40827 * config/i386/avx512vlintrin.h
40828 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
40829 (_mm_mmask_i32gather_ps): Ditto.
40830 (_mm256_mmask_i32gather_pd): Ditto.
40831 (_mm_mmask_i32gather_pd): Ditto.
40832 (_mm256_mmask_i64gather_ps): Ditto.
40833 (_mm_mmask_i64gather_ps): Ditto.
40834 (_mm256_mmask_i64gather_pd): Ditto.
40835 (_mm_mmask_i64gather_pd): Ditto.
40836 (_mm256_mmask_i32gather_epi32): Ditto.
40837 (_mm_mmask_i32gather_epi32): Ditto.
40838 (_mm256_mmask_i32gather_epi64): Ditto.
40839 (_mm_mmask_i32gather_epi64): Ditto.
40840 (_mm256_mmask_i64gather_epi32): Ditto.
40841 (_mm_mmask_i64gather_epi32): Ditto.
40842 (_mm256_mmask_i64gather_epi64): Ditto.
40843 (_mm_mmask_i64gather_epi64): Ditto.
40844 (_mm256_i32scatter_ps): Change __addr type to void*.
40845 (_mm256_mask_i32scatter_ps): Ditto.
40846 (_mm_i32scatter_ps): Ditto.
40847 (_mm_mask_i32scatter_ps): Ditto.
40848 (_mm256_i32scatter_pd): Ditto.
40849 (_mm256_mask_i32scatter_pd): Ditto.
40850 (_mm_i32scatter_pd): Ditto.
40851 (_mm_mask_i32scatter_pd): Ditto.
40852 (_mm256_i64scatter_ps): Ditto.
40853 (_mm256_mask_i64scatter_ps): Ditto.
40854 (_mm_i64scatter_ps): Ditto.
40855 (_mm_mask_i64scatter_ps): Ditto.
40856 (_mm256_i64scatter_pd): Ditto.
40857 (_mm256_mask_i64scatter_pd): Ditto.
40858 (_mm_i64scatter_pd): Ditto.
40859 (_mm_mask_i64scatter_pd): Ditto.
40860 (_mm256_i32scatter_epi32): Ditto.
40861 (_mm256_mask_i32scatter_epi32): Ditto.
40862 (_mm_i32scatter_epi32): Ditto.
40863 (_mm_mask_i32scatter_epi32): Ditto.
40864 (_mm256_i32scatter_epi64): Ditto.
40865 (_mm256_mask_i32scatter_epi64): Ditto.
40866 (_mm_i32scatter_epi64): Ditto.
40867 (_mm_mask_i32scatter_epi64): Ditto.
40868 (_mm256_i64scatter_epi32): Ditto.
40869 (_mm256_mask_i64scatter_epi32): Ditto.
40870 (_mm_i64scatter_epi32): Ditto.
40871 (_mm_mask_i64scatter_epi32): Ditto.
40872 (_mm256_i64scatter_epi64): Ditto.
40873 (_mm256_mask_i64scatter_epi64): Ditto.
40874 (_mm_i64scatter_epi64): Ditto.
40875 (_mm_mask_i64scatter_epi64): Ditto.
40876 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
40877 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
40878 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
40879 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
40880 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
40881 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
40882 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
40883 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
40884 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
40885 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
40886 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
40887 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
40888 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
40889 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
40890 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
40891 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
40892 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
40893 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
40894 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
40895 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
40896 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
40897 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
40898 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
40899 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
40900 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
40901 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
40902 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
40903 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
40904 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
40905 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
40906 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
40907 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
40908 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
40909 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
40910 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
40911 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
40912 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
40913 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
40914 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
40915 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
40916 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
40917 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
40918 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
40919 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
40920 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
40921 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
40922 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
40923 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
40924 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
40925 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
40926 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
40927 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
40928 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
40929 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
40930 definitions accordingly.
40932 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
40933 Kuan-Lin Chen <kuanlinchentw@gmail.com>
40936 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
40939 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
40942 * ira-conflicts.c (ira_build_conflicts): Update total conflict
40943 hard regs for inner regno.
40945 2017-01-17 Martin Liska <mliska@suse.cz>
40948 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
40949 assumption and add comment.
40951 2017-01-17 Nathan Sidwell <nathan@acm.org>
40953 * ipa-visibility.c (localize_node): New function, broken out of ...
40954 (function_and_variable_visibility): ... here. Call it.
40956 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
40958 PR middle-end/77445
40959 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
40960 correctly set frequency of oudgoing edge.
40961 (duplicate_thread_path): Fix profile updating.
40963 2017-01-17 Jakub Jelinek <jakub@redhat.com>
40966 * configure.ac: Add GCC_BASE_VER.
40967 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
40968 version from BASE-VER file.
40969 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
40970 (gcc.o): Depend on $(BASEVER).
40971 * common.opt (dumpfullversion): New option.
40972 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
40973 * doc/invoke.texi: Document -dumpfullversion.
40974 * doc/install.texi: Document --with-gcc-major-version-only.
40975 * configure: Regenerated.
40977 2017-01-17 Richard Biener <rguenther@suse.de>
40979 PR tree-optimization/71433
40980 * tree-vrp.c (register_new_assert_for): Merge same asserts
40981 on all incoming edges.
40982 (process_assert_insertions_for): Handle insertions at the
40985 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
40987 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
40988 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
40990 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
40993 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
40996 2017-01-17 Alan Modra <amodra@gmail.com>
40999 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
41000 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
41001 symbolic stack limit when pic.
41003 2017-01-16 Martin Sebor <msebor@redhat.com>
41005 PR tree-optimization/78608
41006 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
41008 2017-01-16 Jeff Law <law@redhat.com>
41011 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
41012 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
41013 for several include directories that may be relative to sysroot.
41014 * config/i386/x-mingw32 (gplus_includedir): Define.
41015 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
41016 (native_system_includedir): Likewise.
41017 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
41018 override if TARGET_SYSTEM_ROOT is defined.
41019 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
41021 PR tree-optimization/79090
41022 PR tree-optimization/33562
41023 PR tree-optimization/61912
41024 PR tree-optimization/77485
41025 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
41026 and computed trims into the dump file.
41028 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
41030 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
41032 2017-01-16 Jakub Jelinek <jakub@redhat.com>
41035 * gimplify.c (gimplify_init_constructor): If want_value and
41036 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
41040 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
41041 sequence. Formatting fixes.
41042 (doloop_optimize): Formatting fixes.
41045 * gcc.c (debug_level_greater_than_spec_func): New function.
41046 (static_spec_functions): Add debug-level-gt spec function.
41047 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
41049 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
41050 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
41051 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
41052 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
41053 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
41054 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
41056 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
41058 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
41059 QImode fixups to general and mask registers only.
41061 2017-01-16 Carl Love <cel@us.ibm.com>
41063 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
41064 for built-in functions
41065 vector signed char vec_nabs (vector signed char)
41066 vector signed short vec_nabs (vector signed short)
41067 vector signed int vec_nabs (vector signed int)
41068 vector signed long long vec_nabs (vector signed long long)
41069 vector float vec_nabs (vector float)
41070 vector double vec_nabs (vector double)
41071 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
41073 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
41074 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
41075 * doc/extend.texi: Update the documentation file for the new built-in
41078 2017-01-16 Martin Sebor <msebor@redhat.com>
41080 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
41083 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41085 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
41086 UNSPEC_VSX__XXSPLTD to require special splat handling.
41088 2017-01-16 David Malcolm <dmalcolm@redhat.com>
41091 * system.h: Poison strndup.
41093 2017-01-16 Alan Modra <amodra@gmail.com>
41096 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
41099 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
41101 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
41103 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
41105 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
41106 call recog here. Assert that INSN_CODE (insn) is non-negative.
41108 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41111 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
41113 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
41114 in the currently scheduled RTL fragment.
41116 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
41118 PR rtl-optimization/78751
41119 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
41122 2017-01-14 Jeff Law <law@redhat.com>
41124 PR tree-optimization/79090
41125 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
41126 variable length stores.
41127 (compute_trims): Delete dead assignment to *trim_tail.
41128 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
41131 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
41133 PR rtl-optimization/78626
41134 PR rtl-optimization/78727
41135 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
41136 of a block, and split such blocks after everything else is finished.
41138 2017-01-14 Alan Modra <amodra@gmail.com>
41141 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
41142 target legitimate_combined_insn.
41143 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
41144 (rs6000_legitimate_combined_insn): New function.
41145 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
41147 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
41148 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
41149 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
41151 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
41153 * doc/frontends.texi (G++ and GCC): Remove references to Java.
41155 2017-01-13 Jeff Law <law@redhat.com>
41157 PR tree-optimization/33562
41158 PR tree-optimization/61912
41159 PR tree-optimization/77485
41160 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
41162 (delete_dead_assignment): Likewise.
41163 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
41164 statement to delete_dead_call and delete_dead_assignment.
41166 2017-01-13 David Malcolm <dmalcolm@redhat.com>
41169 * substring-locations.c (format_warning_va): Strengthen case 1 so
41170 that both endpoints of the substring must be within the format
41171 range for just the substring to be printed.
41173 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
41175 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
41176 * config/i386/i386.c (ix86_target_string): Add missing options
41177 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
41178 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
41179 flags_other and ix86_target_other to flags2_other. Display unknown
41181 (ix86_valid_target_attribute_inner_p): Add missing options and
41182 reorder options by implied ISAs, as in ix86_target_string.
41184 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41186 * hash-table.h (hash_table::too_empty_p): New function.
41187 (hash_table::expand): Use it.
41188 (hash_table::traverse): Likewise.
41189 (hash_table::empty_slot): Use sizeof (value_type) instead of
41190 sizeof (PTR) to convert bytes to elements. Shrink the table
41191 if the current size is excessive for the current number of
41194 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
41196 * ira-costs.c (record_reg_classes): Break from the inner loop
41197 early once alt_fail is known to be true. Update outer loop
41198 handling accordingly.
41200 2017-01-13 Jeff Law <law@redhat.com>
41202 * tree-ssa-dse.c (decrement_count): New function.
41203 (increment_start_addr, maybe_trim_memstar_call): Likewise.
41204 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
41205 when we know the partially dead statement is a mem* function.
41207 PR tree-optimization/61912
41208 PR tree-optimization/77485
41209 * tree-ssa-dse.c: Include expr.h.
41210 (maybe_trim_constructor_store): New function.
41211 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
41213 PR tree-optimization/33562
41214 PR tree-optimization/61912
41215 PR tree-optimization/77485
41216 * doc/invoke.texi: Document new dse-max-object-size param.
41217 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
41218 * tree-ssa-dse.c: Include params.h.
41219 (dse_store_status): New enum.
41220 (initialize_ao_ref_for_dse): New, partially extracted from
41222 (valid_ao_ref_for_dse, normalize_ref): New.
41223 (setup_live_bytes_from_ref, compute_trims): Likewise.
41224 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
41225 (maybe_trim_partially_dead_store): Likewise.
41226 (maybe_trim_complex_store): Likewise.
41227 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
41228 Track what bytes live from the original store. Return tri-state
41229 for dead, partially dead or live.
41230 (dse_dom_walker): Add constructor, destructor and new private members.
41231 (delete_dead_call, delete_dead_assignment): New extracted from
41233 (dse_optimize_stmt): Make a member of dse_dom_walker.
41234 Use initialize_ao_ref_for_dse.
41236 PR tree-optimization/33562
41237 PR tree-optimization/61912
41238 PR tree-optimization/77485
41239 * sbitmap.h (bitmap_count_bits): Prototype.
41240 (bitmap_clear_range, bitmap_set_range): Likewise.
41241 * sbitmap.c (bitmap_clear_range): New function.
41242 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
41244 2017-01-13 Martin Liska <mliska@suse.cz>
41247 * function.c (set_cfun): Add new argument force.
41248 * function.h (set_cfun): Likewise.
41249 * ipa-inline-transform.c (inline_call): Use the function when
41250 strict alising from is dropped for function we inline to.
41252 2017-01-13 Richard Biener <rguenther@suse.de>
41254 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
41255 for dumping GIMPLE INTEGER_CSTs.
41257 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
41259 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
41260 to 201112L since C++17.
41262 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
41265 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
41266 if -fsanitize=kernel-address is present.
41268 2017-01-13 Richard Biener <rguenther@suse.de>
41270 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
41271 as _Literal ( type ) number in case usual suffixes do not
41272 preserve all information.
41274 2017-01-13 Richard Biener <rguenther@suse.de>
41276 PR tree-optimization/77283
41277 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
41278 and ssa-iterators.h.
41279 (is_feasible_trace): Implement a cost model based on joiner
41282 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
41285 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
41286 char or short to __float128/_Float128 directly.
41288 2017-01-12 Martin Sebor <msebor@redhat.com>
41290 to -Wformat-overflow.
41291 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
41292 (min_bytes_remaining): Same.
41293 (get_string_length): Same.
41294 (format_string): Same.
41295 (format_directive): Same.
41297 (pass_sprintf_length::handle_gimple_call): Same.
41299 2017-01-12 Jakub Jelinek <jakub@redhat.com>
41301 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
41302 info.nowrite calls with no lhs that can't throw. Return bool
41303 whether gsi_remove has been called or not.
41304 (pass_sprintf_length::handle_gimple_call): Return bool whether
41305 try_substitute_return_value called gsi_remove. Formatting fix.
41306 (pass_sprintf_length::execute): Don't use gsi_remove if
41307 handle_gimple_call returned true.
41310 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
41311 be removed due to side-effects, don't remove following barrier nor
41312 turn the successor edge into fallthru edge.
41314 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
41317 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
41318 element-reversing loads and stores as not swappable.
41320 2017-01-12 Nathan Sidwell <nathan@acm.org>
41321 Nicolai Stange <nicstange@gmail.com>
41323 * combine.c (try_combine): Don't ignore result of overlap checking
41324 loop. Combine overlap & asm check into single loop.
41326 2017-01-12 Richard Biener <rguenther@suse.de>
41328 * tree-pretty-print.c (dump_generic_node): Provide -gimple
41329 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
41331 2017-01-12 Richard Biener <rguenther@suse.de>
41333 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
41334 and TS_TARGET_OPTION directly derive from TS_BASE.
41335 * tree-core.h (tree_optimization_option): Derive from tree_base.
41336 (tree_target_option): Likewise.
41338 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
41340 * config/i386/i386.c (memory_address_length): Increase len
41341 only when rip_relative_addr_p returns false.
41343 2017-01-11 Julia Koval <julia.koval@intel.com>
41345 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
41346 (OPTION_MASK_ISA_SGX_SET): New.
41347 (ix86_handle_option): Handle OPT_msgx.
41348 * config.gcc: Added sgxintrin.h.
41349 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
41350 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
41351 * config/i386/i386.c (ix86_target_string): Add -msgx.
41353 (ix86_option_override_internal): Handle new options.
41354 (ix86_valid_target_attribute_inner_p): Add sgx.
41355 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
41356 * config/i386/i386.opt: Add msgx.
41357 * config/i386/sgxintrin.h: New file.
41358 * config/i386/x86intrin.h: Add sgxintrin.h.
41360 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41363 * fold-const.c (maybe_nonzero_address): Return 1 for function
41365 (tree_single_nonzero_warnv_p): Don't handle function local objects
41369 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
41372 2017-01-11 David Malcolm <dmalcolm@redhat.com>
41375 * opts.c: Include "spellcheck.h"
41376 (struct string_fragment): New struct.
41377 (struct edit_distance_traits<const string_fragment &>): New
41379 (get_closest_sanitizer_option): New function.
41380 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
41382 2017-01-11 Jakub Jelinek <jakub@redhat.com>
41384 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
41386 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
41387 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
41388 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
41389 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
41390 for initial die_offset if dwarf_split_debug_info.
41391 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
41392 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
41394 (output_skeleton_debug_sections): Formatting fix. Use
41395 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
41396 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
41398 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
41400 * config/arm/cortex-a53.md: Add bypasses for
41401 cortex_a53_r2f_cvt.
41402 (cortex_a53_r2f): Only use for transfers.
41403 (cortex_a53_f2r): Likewise.
41404 (cortex_a53_r2f_cvt): Add reservation for conversions.
41405 (cortex_a53_f2r_cvt): Likewise.
41407 2017-01-11 Tamar Christina <tamar.christina@arm.com>
41409 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
41410 to all inlined functions, change static to extern.
41412 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
41415 * config/arm/arm.c (legitimize_pic_address): Handle reference to
41417 (arm_assemble_integer): Likewise.
41419 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
41421 * config.gcc: Use new awk script to check CPU, FPU and architecture
41422 parameters for --with-... options.
41423 * config/arm/parsecpu.awk: New file
41424 * config/arm/arm-cpus.in: New file.
41425 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
41427 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
41429 * config/arm/t-arm: Update dependency rules.
41430 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
41431 of processing .def files.
41432 * config/arm/genopt.sh: Deleted.
41433 * config/arm/gentune.sh: Deleted.
41434 * config/arm/arm-cores.def: Deleted.
41435 * config/arm/arm-arches.def: Deleted.
41436 * config/arm/arm-fpus.def: Deleted.
41437 * config/arm/arm-tune.md: Regenerated.
41438 * config/arm/arm-tables.opt: Regenerated.
41439 * config/arm/arm-cpu.h: New generated file.
41440 * config/arm/arm-cpu-data.h: New generated file.
41441 * config/arm/arm-cpu-cdata.h: New generated file.
41443 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
41446 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
41448 (input_varpool_node): Unpack dynamically_initialized bit.
41450 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
41452 PR rtl-optimization/79032
41453 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
41454 the alignment of the adjusted memory reference against that of MODE,
41455 instead of the alignment of the original memory reference.
41457 2017-01-11 Martin Jambor <mjambor@suse.cz>
41459 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
41461 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
41462 decorated functions.
41464 2017-01-11 Richard Biener <rguenther@suse.de>
41466 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
41467 set range/nonnull info for PHI results. Do not set it on
41468 stmts marked for removal.
41470 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
41472 * expr.c (store_field): In the bitfield case, fetch the return value
41473 from the registers before applying a single big-endian adjustment.
41474 Always do a final load for a BLKmode value not larger than a word.
41476 2017-01-10 David Malcolm <dmalcolm@redhat.com>
41479 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41480 that we correctly handle column numbers greater than
41481 LINE_MAP_MAX_COLUMN_NUMBER.
41483 2017-01-10 Martin Sebor <msebor@redhat.com>
41485 PR middle-end/78245
41486 * gimple-ssa-sprintf.c (get_destination_size): Call
41487 {init,fini}object_sizes.
41488 * tree-object-size.c (addr_object_size): Adjust.
41489 (pass_through_call): Adjust.
41490 (pass_object_sizes::execute): Adjust.
41491 * tree-object-size.h (fini_object_sizes): Declare.
41493 2017-01-10 Martin Sebor <msebor@redhat.com>
41495 PR tree-optimization/78775
41496 * builtins.c (get_size_range): Move...
41497 * calls.c: ...to here.
41498 (alloc_max_size): Accept zero argument.
41499 (operand_signed_p): Remove.
41500 (maybe_warn_alloc_args_overflow): Call get_size_range.
41501 * calls.h (get_size_range): Declare.
41503 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
41505 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
41506 from TI's devices.csv file as of September 2016.
41507 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
41509 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
41511 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
41512 * doc/invoke.texi: Likewise.
41513 * doc/md.texi: Likewise.
41514 * doc/objc.texi: Likewise.
41516 2017-01-10 Joshua Conner <joshconner@google.com>
41518 * config/arm/fuchsia-elf.h: New file.
41519 * config/fuchsia.h: New file.
41520 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
41521 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
41523 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
41525 2016-01-10 Richard Biener <rguenther@suse.de>
41527 PR tree-optimization/79034
41528 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
41529 Propagate out degenerate PHIs in the joiner.
41531 2017-01-10 Martin Liska <mliska@suse.cz>
41533 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
41534 (sort_congruence_classes_by_decl_uid): Likewise.
41535 (sort_congruence_class_groups_by_decl_uid): Likewise.
41536 (sem_item_optimizer::merge_classes): Sort class, groups in these
41537 classes and members in the groups by DECL_UID of declarations.
41538 This would make merge operations stable.
41540 2017-01-10 Martin Liska <mliska@suse.cz>
41542 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
41543 usage of m_classes_vec.
41544 (sem_item_optimizer::~sem_item_optimizer): Likewise.
41545 (sem_item_optimizer::get_group_by_hash): Likewise.
41546 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
41547 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
41548 (sem_item_optimizer::verify_classes): Likewise.
41549 (sem_item_optimizer::process_cong_reduction): Likewise.
41550 (sem_item_optimizer::dump_cong_classes): Likewise.
41551 (sem_item_optimizer::merge_classes): Likewise.
41552 * ipa-icf.h (congruence_class_hash): Rename from
41553 congruence_class_group_hash. Remove declaration of m_classes_vec.
41555 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
41557 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
41558 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
41559 * config.gcc: Add avx512vpopcntdqintrin.h.
41560 * config/i386/avx512vpopcntdqintrin.h: New.
41561 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
41562 * config/i386/i386-builtin-types.def: Add new types.
41563 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
41564 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
41565 __builtin_ia32_vpopcountq_v8di_mask): New.
41566 * config/i386/i386-c.c (ix86_target_macros_internal): Define
41567 __AVX512VPOPCNTDQ__.
41568 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
41569 (PTA_AVX512VPOPCNTDQ): Define.
41570 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
41571 TARGET_AVX512VPOPCNTDQ_P): Define.
41572 * config/i386/i386.opt: Add mavx512vpopcntdq.
41573 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
41574 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
41576 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41578 PR middle-end/77484
41579 * predict.def (PRED_CALL): Set to 67.
41581 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
41583 * expr.c (store_field): In the bitfield case, if the value comes from
41584 a function call and is of an aggregate type returned in registers, do
41585 not modify the field mode; extract the value in all cases if the mode
41586 is BLKmode and the size is not larger than a word.
41588 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
41591 * config/i386/cpuid.h: Fix undefined behavior.
41593 2017-01-04 Jeff Law <law@redhat.com>
41595 PR tree-optimization/79007
41596 PR tree-optimization/67955
41597 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
41598 conservative for pt.null when flag_non_call_exceptions is on.
41600 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41602 PR translation/79019
41603 PR translation/79020
41604 * params.def (PARAM_INLINE_MIN_SPEEDUP,
41605 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
41606 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
41608 * config/avr/avr.opt (maccumulate-args): Likewise.
41609 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
41610 * common.opt (freport-bug): Likewise.
41611 * cif-code.def (CIF_FINAL_ERROR): Likewise.
41612 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
41613 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
41614 translatable string.
41615 * config/i386/i386.c (function_value_32): Likewise.
41616 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
41617 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
41619 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
41620 * common/config/msp430/msp430-common.c (msp430_handle_option):
41622 * symtab.c (symtab_node::verify_base): Likewise.
41623 * opts.c (set_debug_level): Likewise.
41624 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
41625 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
41626 missing whitespace to translatable strings.
41627 * config/avr/avr.md (bswapsi2): Fix typo in comment.
41628 * config/sh/superh.h: Likewise.
41629 * config/i386/xopintrin.h: Likewise.
41630 * config/i386/znver1.md: Likewise.
41631 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
41632 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
41633 * double-int.h (struct double_int): Likewise.
41634 * double-int.c (div_and_round_double): Likewise.
41635 * wide-int.cc: Likewise.
41636 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
41637 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
41638 * cfgcleanup.c (crossjumps_occured): Renamed to ...
41639 (crossjumps_occurred): ... this.
41640 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
41643 PR tree-optimization/78899
41644 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
41645 returning bool return struct loop *, NULL for failure and the new
41647 (versionable_outer_loop_p): Don't version outer loop if it has
41648 dont_vectorized bit set.
41649 (tree_if_conversion): When versioning outer loop, ensure
41650 tree_if_conversion is performed also on the inner loop of the
41651 non-vectorizable outer loop copy.
41652 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
41653 LOOP_VECTORIZED in inner loop of the scalar outer loop and
41654 prevent vectorization of it.
41655 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
41656 the outer loop vectorization of the non-scalar version is attempted
41657 before vectorization of the inner loop in scalar version. If
41658 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
41659 vectorization of its inner loop.
41660 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
41661 has 2 inner loops, rename also on edges from bb whose single pred
41662 is outer_loop->header. Fix typo in function comment.
41664 2017-01-09 Martin Sebor <msebor@redhat.com>
41667 * asan.c (asan_emit_stack_protection): Increase local buffer size
41668 to avoid snprintf truncation warning.
41670 2017-01-09 Andrew Pinski <apinski@cavium.com>
41672 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
41673 to reference thunderx2t99 for the tuning structure
41674 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
41676 (thunderx2t99_extra_costs): This.
41677 * config/aarch64/aarch64-tune.md: Regenerate.
41678 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
41679 (vulcan_addrcost_table): This.
41680 (vulcan_regmove_cost): Rename to ...
41681 (thunderx2t99_regmove_cost): This.
41682 (vulcan_vector_cost): Rename to ...
41683 (thunderx2t99_vector_cost): this.
41684 (vulcan_branch_cost): Rename to ...
41685 (thunderx2t99_branch_cost): This.
41686 (vulcan_tunings): Rename to ...
41687 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
41688 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
41690 2017-01-09 Martin Jambor <mjambor@suse.cz>
41694 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
41695 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
41696 (propagate_vr_accross_jump_function): Use the above function for all
41697 value range computations for pass-through jump functions and type
41698 converasion from explicit value range values.
41699 (ipcp_propagate_stage): Do not attempt to deduce types of formal
41700 parameters from TYPE_ARG_TYPES.
41701 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
41702 (ipa_write_node_info): Stream type of the actual argument.
41703 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
41705 2017-01-09 Martin Liska <mliska@suse.cz>
41708 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
41709 (lookup_compiler): Do not show error message with have_E.
41711 2017-01-09 Jakub Jelinek <jakub@redhat.com>
41713 PR tree-optimization/78938
41714 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
41715 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
41716 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
41717 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
41720 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
41722 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
41725 2017-01-09 Richard Biener <rguenther@suse.de>
41727 PR tree-optimization/78997
41728 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
41729 name condition properly.
41731 2017-01-09 Richard Biener <rguenther@suse.de>
41734 * dwarf2out.c (is_cxx): New overload with context.
41735 (is_naming_typedef_decl): Use it.
41737 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41739 * invoke.texi (Option Summary): Correct spacing in option lists
41740 and add line breaks to fix over-long lines.
41742 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41744 PR middle-end/17660
41746 * extend.texi (Common Variable Attributes): Add xref to GCC
41747 Internals manual to explain mode attribute keywords.
41749 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
41752 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
41753 and Preprocessor Options.
41754 (Options for Linking): Document -pthread here....
41755 (RS/6000 and PowerPC Options): ...not here.
41756 (Solaris 2 Options): ...or here.
41757 * doc/cppopts.texi: Document -pthread.
41759 2017-01-08 Martin Sebor <msebor@redhat.com>
41761 PR middle-end/77708
41762 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
41763 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
41764 New member functions.
41765 (format_directive): Used them.
41767 (pass_sprintf_length::handle_gimple_call): Same.
41768 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
41769 to avoid truncation for any argument.
41770 (extract_affine_mul): Same.
41771 * tree.c (get_file_function_name): Same.
41773 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
41775 PR middle-end/77484
41776 * predict.def (PRED_INDIR_CALL): Set to 86.
41778 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41780 PR preprocessor/54124
41781 * doc/cppopts.texi: Reformat -d subtable to list the full name
41782 of the options. Add cross-reference to the docs for the general
41783 compiler -d options.
41784 * doc/invoke.texi (Developer Options): Add cross-reference to the
41785 preprocessor-specific -d option documentation.
41787 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41789 PR preprocessor/13498
41790 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
41791 redudant material, and reflect new command-line options.
41792 (System Headers): Likewise.
41794 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
41796 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
41797 -isystem, and -idirafter. Copy-edit.
41798 * doc/cppopts.texi: Copy-edit. Remove contradiction about
41799 default for -ftrack-macro-expansion. Delete obsolete and
41800 badly-formatted implementation details about -fdebug-cpp output.
41801 * doc/cppwarnopts.texi: Copy-edit.
41803 2017-01-07 David Malcolm <dmalcolm@redhat.com>
41806 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
41807 that the transition from a max line width >= 1<<10 to narrower
41808 lines works correctly.
41810 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
41812 * doc/options.texi (PerFunction): New.
41813 * opt-functions.awk (switch_flags): Map both Optimization and
41814 PerFunction to CL_OPTIMIZATION.
41815 * opth-gen.awk: Test for PerFunction flag along with
41817 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
41818 it only when the latter is present. Skip those that don't in
41819 the hash function generator.
41820 * common.opt (fvar-tracking): Mark as PerFunction instead of
41822 (fvar-tracking-assignments): Likewise.
41823 (fvar-tracking-assignments-toggle): Likewise.
41824 (fvar-tracking-uninit): Likewise.
41826 2017-01-07 Jakub Jelinek <jakub@redhat.com>
41828 PR translation/79018
41829 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
41832 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
41835 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
41836 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
41837 TARGET_LONG_JUMP_TABLE_OFFSETS.
41838 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
41839 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
41840 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
41841 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
41842 * config/m68k/m68k.md (tablejump expander): Likewise.
41843 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
41844 TARGET_LONG_JUMP_TABLE_OFFSETS.
41845 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
41846 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
41848 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
41849 David Holsgrove <david.holsgrove@xilinx.com>
41851 * common/config/microblaze/microblaze-common.c
41852 (TARGET_EXCEPT_UNWIND_INFO): Remove.
41853 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
41855 * config/microblaze/microblaze.c (microblaze_must_save_register)
41856 (microblaze_expand_epilogue, microblaze_return_addr): Handle
41858 (microblaze_eh_return): New function.
41859 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
41860 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
41861 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
41862 * config/microblaze/microblaze.md (eh_return): New pattern.
41864 2017-01-06 Jakub Jelinek <jakub@redhat.com>
41866 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
41867 GCC_DIAGNOSTIC_STRINGIFY): Define.
41869 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
41871 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41873 * config/arm/arm.md (<mcrr>): New.
41875 * config/arm/arm.c (arm_arch5te): New.
41876 (arm_option_override): Set arm_arch5te.
41877 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
41879 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
41880 (arm_mcrr_qualifiers): ... this. New.
41881 (MRRC_QUALIFIERS): Define to...
41882 (arm_mrrc_qualifiers): ... this. New.
41883 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
41885 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
41886 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
41887 (MRRCI, mrrc, MRRC): New.
41888 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
41889 VUNSPEC_MRRC2): New.
41891 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41893 * config/arm/arm.md (<mcr>): New.
41895 * config/arm/arm.c (arm_coproc_builtin_available): Add
41896 support for mcr, mrc, mcr2 and mrc2.
41897 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
41898 (arm_mcr_qualifiers): ... this. New.
41899 (MRC_QUALIFIERS): Define to ...
41900 (arm_mrc_qualifiers): ... this. New.
41901 (MCR_QUALIFIERS): Define to ...
41902 (arm_mcr_qualifiers): ... this. New.
41903 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
41905 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
41906 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
41907 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
41908 VUNSPEC_MRC2): New.
41910 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41912 * config/arm/arm.md (*ldc): New.
41916 * config/arm/arm.c (arm_coproc_builtin_available): Add
41917 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
41918 (arm_coproc_ldc_stc_legitimate_address): New.
41919 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
41920 'qualifier_const_pointer'.
41921 (LDC_QUALIFIERS): Define to...
41922 (arm_ldc_qualifiers): ... this. New.
41923 (STC_QUALIFIERS): Define to...
41924 (arm_stc_qualifiers): ... this. New.
41925 * config/arm/arm-protos.h
41926 (arm_coproc_ldc_stc_legitimate_address): New.
41927 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
41928 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
41929 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
41930 stc2, stcl, stc2l): New.
41931 * config/arm/constraints.md (Uz): New.
41932 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
41933 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
41934 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
41935 VUNSPEC_STC2L): New.
41937 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41939 * config/arm/arm.md (<cdp>): New.
41940 * config/arm/arm.c (neon_const_bounds): Rename this ...
41941 (arm_const_bounds): ... this.
41942 (arm_coproc_builtin_available): New.
41943 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
41944 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
41945 (CDP_QUALIFIERS): Define to...
41946 (arm_cdp_qualifiers): ... this. New.
41948 (arm_expand_builtin_args): Add case for 6 arguments.
41949 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
41950 (arm_const_bounds): ... this.
41951 (arm_coproc_builtin_available): New.
41952 * config/arm/arm_acle.h (__arm_cdp): New.
41954 * config/arm/arm_acle_builtins.def (cdp): New.
41956 * config/arm/iterators.md (CDPI,CDP,cdp): New.
41957 * config/arm/neon.md: Rename all 'neon_const_bounds' to
41958 'arm_const_bounds'.
41959 * config/arm/types.md (coproc): New.
41960 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
41961 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
41962 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
41963 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
41965 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41967 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
41968 (UBINOP_QUALIFIERS): New.
41970 (acle_builtin_data): New. Change comment.
41971 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
41972 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
41973 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
41974 arm_acle_builtins.def.
41975 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
41976 (arm_init_acle_builtins): New.
41977 (CRC32_BUILTIN): Remove.
41978 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
41979 crc32cb, crc32ch and crc32cw.
41980 (arm_init_crc32_builtins): Remove.
41981 (arm_init_builtins): Use arm_init_acle_builtins rather
41982 than arm_init_crc32_builtins.
41983 (arm_expand_acle_builtin): New.
41984 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
41985 * config/arm/arm_acle_builtins.def: New.
41987 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
41989 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
41990 (arm_builtin_datum): ... this.
41991 (arm_init_neon_builtin): Rename to ...
41992 (arm_init_builtin): ... this. Add a new parameters PREFIX
41993 and USE_SIG_IN_NAME.
41994 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
41995 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
41996 'arm_builtin_datum'.
41997 (arm_init_vfp_builtins): Likewise.
41998 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
41999 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
42000 (arm_expand_neon_args): Rename to ...
42001 (arm_expand_builtin_args): ... this. Rename builtin_arg
42002 enum values and differentiate between ARG_BUILTIN_MEMORY
42003 and ARG_BUILTIN_NEON_MEMORY.
42004 (arm_expand_neon_builtin_1): Rename to ...
42005 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
42006 values, arm_expand_builtin_args and add bool parameter NEON.
42007 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
42008 (arm_expand_vfp_builtin): Likewise.
42009 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
42011 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42013 PR middle-end/77484
42014 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
42015 * predict.c (tree_estimate_probability_bb): Reverse direction of
42016 polymorphic call predictor.
42018 2017-01-06 David Malcolm <dmalcolm@redhat.com>
42020 * passes.c (execute_one_pass): Split out pass-skipping logic into...
42021 (determine_pass_name_match): ...this new function and...
42022 (should_skip_pass_p): ...this new function.
42024 2017-01-06 Nathan Sidwell <nathan@acm.org>
42026 * ipa-visibility.c (function_and_variable_visibility): Reformat
42027 comments and long lines. Remove extrneous if.
42028 * symtab.c (symtab_node::make_decl_local): Fix code format.
42029 (symtab_node::set_section_for_node): Fix comment typo.
42031 2017-01-06 Martin Liska <mliska@suse.cz>
42034 * lra-constraints.c: Rename invariant to lra_invariant.
42035 * predict.c (set_even_probabilities): Initialize e to NULL.
42037 2017-01-05 Martin Sebor <msebor@redhat.com>
42039 PR tree-optimization/78910
42040 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
42041 (format_integer): Correct off-by-one error in the handling
42042 of precision with negative numbers in signed conversions..
42044 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
42046 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
42048 2017-01-05 Jakub Jelinek <jakub@redhat.com>
42050 PR tree-optimization/71016
42051 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
42052 factor_out_conditional_conversion. Formatting fix.
42053 (factor_out_conditional_conversion): Add cond_stmt argument.
42054 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
42055 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
42058 2017-01-05 David Malcolm <dmalcolm@redhat.com>
42060 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
42061 read-rtl-function.o, and selftest-rtl.o.
42062 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
42063 (selftest::aarch64_test_loading_full_dump): New function.
42064 (selftest::aarch64_run_selftests): New function.
42065 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
42066 selftest::aarch64_run_selftests.
42067 * config/i386/i386.c
42068 (selftest::ix86_test_loading_dump_fragment_1): New function.
42069 (selftest::ix86_test_loading_call_insn): New function.
42070 (selftest::ix86_test_loading_full_dump): New function.
42071 (selftest::ix86_test_loading_unspec): New function.
42072 (selftest::ix86_run_selftests): Call the new functions.
42073 * emit-rtl.c (maybe_set_max_label_num): New function.
42074 * emit-rtl.h (maybe_set_max_label_num): New decl.
42075 * function.c (instantiate_decls): Guard call to
42076 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
42077 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
42079 * gensupport.c (gen_reader::gen_reader): Pass "false"
42080 for new "compact" param of rtx_reader.
42081 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
42082 rather than an empty string for NULL strings.
42083 * read-md.c: Potentially include config.h rather than bconfig.h.
42084 Wrap include of errors.h with #ifdef GENERATOR_FILE.
42085 (have_error): New global, copied from errors.c.
42086 (md_reader::read_name): Rename to...
42087 (md_reader::read_name_1): ...this, adding "out_loc" param,
42088 and converting "missing name or number" to returning false, rather
42090 (md_reader::read_name): Reimplement in terms of read_name_1.
42091 (md_reader::read_name_or_nil): New function.
42092 (md_reader::read_string): Handle "(nil)" by returning NULL.
42093 (md_reader::md_reader): Add new param "compact".
42094 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
42095 (md_reader::read_file): New method.
42096 * read-md.h (md_reader::md_reader): Add new param "compact".
42097 (md_reader::read_file): New method.
42098 (md_reader::is_compact): New accessor.
42099 (md_reader::read_name): Convert return type from void to file_location.
42100 (md_reader::read_name_or_nil): New decl.
42101 (md_reader::read_name_1): New decl.
42102 (md_reader::m_compact): New field.
42103 (noop_reader::noop_reader): Pass "false" for new "compact" param
42105 (rtx_reader::rtx_reader): Add new "compact" param.
42106 (rtx_reader::read_rtx_operand): Make virtual and convert return
42107 type from void to rtx.
42108 (rtx_reader::read_until): New decl.
42109 (rtx_reader::handle_any_trailing_information): New virtual function.
42110 (rtx_reader::postprocess): New virtual function.
42111 (rtx_reader::finalize_string): New virtual function.
42112 (rtx_reader::m_in_call_function_usage): New field.
42113 (rtx_reader::m_reuse_rtx_by_id): New field.
42114 * read-rtl-function.c: New file.
42115 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
42116 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
42117 (selftest::verify_three_block_rtl_cfg): New decl.
42118 * read-rtl-function.h: New file.
42119 * read-rtl.c: Potentially include config.h rather than bconfig.h.
42120 For host, include function.h, memmodel.h, and emit-rtl.h.
42121 (one_time_initialization): New function.
42122 (struct compact_insn_name): New struct.
42123 (compact_insn_names): New array.
42124 (find_code): Handle insn codes in compact dumps.
42125 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
42126 (bind_subst_iter_and_attr): Likewise.
42127 (add_condition_to_string): Likewise.
42128 (add_condition_to_rtx): Likewise.
42129 (apply_attribute_uses): Likewise.
42130 (add_current_iterators): Likewise.
42131 (apply_iterators): Likewise.
42132 (initialize_iterators): Guard usage of apply_subst_iterator with
42133 #ifdef GENERATOR_FILE.
42134 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
42135 (md_reader::read_mapping): Likewise.
42136 (add_define_attr_for_define_subst): Likewise.
42137 (add_define_subst_attr): Likewise.
42138 (read_subst_mapping): Likewise.
42139 (check_code_iterator): Likewise.
42140 (rtx_reader::read_rtx): Likewise. Move one-time initialization
42142 (one_time_initialization): New function.
42143 (rtx_reader::read_until): New method.
42144 (read_flags): New function.
42145 (parse_reg_note_name): New function.
42146 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
42147 Handle reuse_rtx ids.
42148 Wrap iterator lookup within #ifdef GENERATOR_FILE.
42149 Add parsing support for RTL dumps, mirroring the special-cases in
42150 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
42151 values, and calling handle_any_trailing_information.
42152 (rtx_reader::read_rtx_operand): Convert return type from void
42153 to rtx, returning return_rtx. Handle case 'e'. Call
42154 finalize_string on XSTR and XTMPL fields.
42155 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
42156 "(nil)" values were omitted. Call the postprocess vfunc on the
42158 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
42159 class ctor. Initialize m_in_call_function_usage. Call
42160 one_time_initialization.
42161 * rtl-tests.c (selftest::test_uncond_jump): Call
42162 set_new_first_and_last_insn.
42163 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
42164 * selftest-rtl.c: New file.
42165 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
42166 (selftest::get_insn_by_uid): New decl.
42167 * selftest-run-tests.c (selftest::run_tests): Call
42168 read_rtl_function_c_tests.
42169 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
42170 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
42173 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
42175 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
42176 operands in a special way. Assert that pos+len <= mode precision.
42178 2017-01-05 Jakub Jelinek <jakub@redhat.com>
42180 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
42181 3 argument Alias with unlimited for the negative form.
42182 (fno-vect-cost-model): Removed.
42184 2017-01-05 Martin Liska <mliska@suse.cz>
42186 * hsa-gen.c (gen_hsa_divmod): New function.
42187 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
42189 2017-01-05 Martin Liska <mliska@suse.cz>
42192 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
42195 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42197 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
42198 small constant length operands.
42200 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
42202 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
42203 between loop iterations.
42205 2017-01-05 Martin Liska <mliska@suse.cz>
42208 * gimplify.c (gimplify_decl_expr): Compare to
42209 asan_poisoned_variables instread of checking flags.
42210 (gimplify_target_expr): Likewise.
42211 (gimplify_expr): Likewise.
42212 (gimplify_function_tree): Conditionally initialize
42213 asan_poisoned_variables.
42215 2017-01-04 Jeff Law <law@redhat.com>
42217 PR tree-optimizatin/78812
42218 * rtl.h (contains_mem_rtx_p): Prototype.
42219 * ifcvt.c (containts_mem_rtx_p): Move from here to...
42220 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
42221 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
42222 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
42223 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
42225 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42227 * input.c (assert_char_at_range): Default-initialize actual_range.
42229 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42231 * df-scan.c (df_ref_create_structure): Make regno unsigned,
42232 to match the caller.
42234 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42236 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
42237 insns after final jump in test to emit dummy move.
42239 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42241 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
42242 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
42244 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
42246 * multiple_target.c (create_dispatcher_calls): Init e_next.
42247 * tree-ssa-loop-split.c (split_loop): Init border.
42248 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
42251 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
42256 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
42257 (altivec_register_operand): Do not return true if the operand
42258 contains a SUBREG mixing SImode and SFmode.
42259 (vsx_register_operand): Likewise.
42260 (vsx_reg_sfsubreg_ok): New predicate.
42261 (vfloat_operand): Do not return true if the operand contains a
42262 SUBREG mixing SImode and SFmode.
42263 (vint_operand): Likewise.
42264 (vlogical_operand): Likewise.
42265 (gpc_reg_operand): Likewise.
42266 (int_reg_operand): Likewise.
42267 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
42268 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
42269 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
42271 (rs6000_emit_move_si_sf_subreg): New helper function.
42272 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
42273 fixup SUBREGs involving SImode and SFmode.
42274 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
42275 numbers for the new peephole2 optimization.
42276 (peephole2 for SFmode unions): New peephole2 to optimize cases in
42277 the GLIBC math library that do AND/IOR/XOR operations on single
42278 precision floating point.
42279 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
42280 target macros to say whether we need to avoid SUBREGs mixing
42282 (TARGET_ALLOW_SF_SUBREG): Likewise.
42283 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
42284 (UNSPEC_SI_FROM_SF): Likewise.
42285 (iorxor): Change spacing.
42286 (and_ior_xor): New iterator for AND, IOR, and XOR.
42287 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
42288 (movdi_from_sf_zero_ext): Likewise.
42289 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
42290 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
42291 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
42292 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
42293 (fms<mode>4): Likewise.
42294 (fnma<mode>4): Likewise.
42295 (fnms<mode>4): Likewise.
42296 (nfma<mode>4): Likewise.
42297 (nfms<mode>4): Likewise.
42299 2017-01-04 Marek Polacek <polacek@redhat.com>
42302 * doc/invoke.texi: Document -Wpointer-compare.
42304 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42306 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
42309 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
42310 descriptions for -gdwarf-5 and emit them as uleb128 instead of
42313 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
42316 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
42317 documentation of the powerpc_popcntb_ok attribute.
42318 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
42319 code to issue warning messages if a requested CPU configuration is
42320 not supported by the binary (assembler and loader) toolchain.
42321 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
42322 made to define a built-in function that has been disabled.
42323 (paired_init_builtins): Add assertion to prevent ICE if attempt is
42324 made to define a built-in function that has been disabled.
42325 (altivec_init_builtins): Add comment explaining why definition
42326 of the DST built-in functions is not preceded by an assertion
42327 check. Add assertions to prevent ICE if attempts are made to
42328 define an altivec predicate or an abs* built-in function that has
42330 (htm_init_builtins): Add comment explaining why definition of the
42331 htm built-in functions is not preceded by an assertion check.
42333 2017-01-04 Jeff Law <law@redhat.com>
42335 PR tree-optimizatin/67955
42336 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
42337 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
42338 the points-to solution does not include pt_null. Use DECL_PT_UID
42341 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
42343 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
42344 Use gen_int_mode instead of gen_lopwart for const_int operands.
42346 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42348 PR tree-optimization/71563
42349 * match.pd: Simplify X << Y into X if Y is known to be 0 or
42350 out of range value - has low bits known to be zero.
42352 2017-01-04 Alan Modra <amodra@gmail.com>
42354 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
42355 * configure: Regenerate.
42356 * config.in: Regenerate.
42358 2017-01-04 Jakub Jelinek <jakub@redhat.com>
42361 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
42362 a substring of the message, but strcmp with the whole message. Ifdef
42363 ENABLE_NLS, translate the message first using dgettext.
42365 2017-01-03 Jeff Law <law@redhat.com>
42367 PR tree-optimizatin/78856
42368 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
42369 (mark_threaded_blocks): Remove code to truncate thread paths that
42370 cross multiple loop headers. Instead invalidate the cached loop
42371 iteration information and handle case of a thread path walking
42372 into an irreducible region.
42374 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
42377 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
42378 assertions. Add support for doing the signbit if the IEEE 128-bit
42379 floating point value is in a GPR.
42380 * config/rs6000/rs6000.md (Fsignbit): Delete.
42381 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
42382 Update the length attribute if the value is in a GPR.
42383 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
42384 the sign or zero extension instruction, since the value is always 0/1.
42385 (signbit<mode>2_dm2): Delete using <Fsignbit>.
42388 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
42389 extracting SImode to a GPR register so that we can generate a
42390 store, limit the vector to be in a traditional Altivec register
42391 for the vextuwrx instruction.
42393 2017-01-03 Ian Lance Taylor <iant@google.com>
42395 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
42397 2017-01-03 Martin Sebor <msebor@redhat.com>
42399 PR tree-optimization/78696
42400 * gimple-ssa-sprintf.c (format_floating): Correct handling of
42401 precision. Use MPFR for %f for greater fidelity. Correct handling
42403 (pass_sprintf_length::compute_format_length): Set width and precision
42404 specified by asrerisk to void_node for vararg functions.
42405 (try_substitute_return_value): Adjust dump output.
42407 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
42409 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
42411 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
42413 * doc/invoke.texi (SPARC options): Document -mlra as the default.
42414 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
42415 -mlra/-mno-lra was passed to the compiler.
42417 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
42419 PR rtl-optimization/65618
42420 * emit-rtl.c (try_split): Move initialization of "before" and
42421 "after" to just before the call to emit_insn_after_setloc.
42423 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
42425 * doc/md.texi (Standard Names): Remove reference to Java frontend.
42427 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
42429 * dwarf2out.c (gen_enumeration_type_die): When
42430 -gno-strict-dwarf, add a DW_AT_encoding attribute.
42432 2017-01-03 Jakub Jelinek <jakub@redhat.com>
42434 PR tree-optimization/78965
42435 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
42436 Change first argument from const call_info & to call_info &. For %n
42437 set info.nowrite to false.
42439 PR middle-end/78901
42440 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
42441 possibly throwing calls.
42443 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
42444 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
42445 and fns handling, rather than in a separate case SSA_NAME.
42447 2017-01-02 Jeff Law <law@redhat.com>
42449 * config/darwin-driver.c (darwin_driver_init): Const-correctness
42450 fixes for first_period and second_period variables.
42452 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
42455 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
42456 (*insvqi_1): New insn pattern.
42457 (*insvqi_1_mem_rex64): Ditto.
42458 (*insvqi_2): Ditto.
42459 (*insvqi_3): Rename from *insvqi.
42461 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
42463 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
42465 * doc/cfg.texi (Edges): Remove reference to Java.
42466 (Maintaining the CFG): Ditto.
42468 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42470 PR middle-end/77674
42471 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
42472 transparent aliases.
42474 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
42476 PR middle-end/77484
42477 * predict.def (PRED_CALL): Update hitrate.
42478 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
42479 * predict.c (tree_estimate_probability_bb): Split CALL predictor
42480 into direct/indirect/polymorphic variants.
42482 2017-01-01 Jakub Jelinek <jakub@redhat.com>
42484 Update copyright years.
42486 * gcc.c (process_command): Update copyright notice dates.
42487 * gcov-dump.c (print_version): Ditto.
42488 * gcov.c (print_version): Ditto.
42489 * gcov-tool.c (print_version): Ditto.
42490 * gengtype.c (create_file): Ditto.
42491 * doc/cpp.texi: Bump @copying's copyright year.
42492 * doc/cppinternals.texi: Ditto.
42493 * doc/gcc.texi: Ditto.
42494 * doc/gccint.texi: Ditto.
42495 * doc/gcov.texi: Ditto.
42496 * doc/install.texi: Ditto.
42497 * doc/invoke.texi: Ditto.
42499 Copyright (C) 2017 Free Software Foundation, Inc.
42501 Copying and distribution of this file, with or without modification,
42502 are permitted in any medium without royalty provided the copyright
42503 notice and this notice are preserved.